Time, Reality &

Experience
ROYAL INSTITUTE OF PHILOSOPHY SUPPLEMENT: 50

EDITED BY

Craig Callender

CAMBRIDGE
UNIVERSITY PRESS
PUBLISHED BY THE PRESS SYNDICATE OF THE UNIVERSITY OF CAMBRIDGE
The Pitt Building, Trumpington Street, Cambridge, CB2 1RP,
United Kingdom

CAMBRIDGE UNIVERSITY PRESS

The Edinburgh Building, Cambridge CB2 2RU, United Kingdom
40 West 20th Street, New York, NY 10011-4211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia

© The Royal Institute of Philosophy and the contributors 2002

Printed in the United Kingdom at the University Press, Cambridge

Typeset by Michael Heath Ltd, Reigate, Surrey

A catalogue record for this book is available

from the British Library

Library of Congress Cataloguing-in-Publication Data

Time, reality & experience/edited by Craig Callender.

p. cm.—(Royal Institute of Philosophy supplement, ISSN
1358-2461; 50)
Includes bibliographical references and index
ISBN 0-521-52967 0 (pbk.)
1. Time—Philosophy. I. Title: Time, reality, and experience. II.
Callender, Craig, 1968. III. Series.
BD638.T566 2002
115—dc21 2002067445

ISBN 0 521 52967 0 paperback

ISSN 1358-2461
Contents

Preface v

Notes on Contributors vii

When Time Gets Off Track 1

JAN FAYE

Burbury's Last Case: The Mystery of the Entropic Arrow 19

HUW PRICE

Zeno's Arrow and the Significance of the Present 57

ROBIN LEPOIDEVIN

Presentism, Ontology and Temporal Experience 73

L. NATHAN OAKLANDER

A Presentist's Refutation of Mellor's McTaggart 91

PHILIP PERCIVAL

Time and Degrees of Existence: A Theory of 'Degree

Presentism' 119
QUENTIN SMITH

McTaggart and the Truth about Time 137

HEATHER DYKE

On Absolute Becoming and the Myth of Passage 153

STEVEN F. SAVITT

Time Travel and Modern Physics 169

FRANK ARNTZENIUS AND TIM MAUDLIN

Freedom from the Inside Out 201

CARL HOEFER

On Stages, Worms and Relativity 223

YURIBALASHOV

in
Contents

On Becoming, Cosmic Time and Rotating Universes 253

MAURO DORATO

How Relativity Contradicts Presentism 277

SIMON SAUNDERS

Can Physics Coherently Deny the Reality of Time? 293

RICHARD HEALEY

Rememberances, Mementos, and Time-Capsules 317

JENANN ISMAEL

IV
Preface
In the past decade the philosophy of time seems to have experi-
enced a renaissance across all of its various sub-fields. With this
activity in mind, the Royal Institute of Philosophy agreed to let me
organize its annual conference around the title, Time, Reality and
Experience. Held at the London School of Economics on 27-8
Sept 2000, the conference brought together philosophers and scien-
tists interested in time from all over the world. The conference was
truly international, with speakers coming from nine different coun-
tries across three different continents. In many cases, prominent
philosophers of time met each other for the first time. The confer-
ence was also wide-ranging in terms of its approach to philosophy
of time, balanced between studies of time as it is found in science
and more traditional analytic philosophy of time. In all, I felt it a
great success.
The London conference was a cause of this book, but it was not
the source of all that is in it. Professor Anthony O'Hear of the
Royal Institute charged me with producing a good collection in
philosophy of time, regardless of the origin of the papers. To round
out the volume, I then searched for more good unpublished papers
in philosophy of time—and found them. I heard two at a wonderful
conference (Real Time and its Quantum Roots, 12-14 April,
2001) organized by Richard Healey and held at the University of
Arizona. They were papers by Richard himself and Jenann Ismael
on the—dare I say 'timely'—topic of timelessness in quantum grav-
ity. Then L. Nathan Oaklander kindly brought to my attention a
paper by Philip Percival, who was an active member of the audience
at the London conference. The last paper I found while browsing
on the internet. There I happened upon an outstanding contribu-
tion on time travel by Frank Arntzenius and Tim Maudlin. Though
already published online, I felt it also deserved a home on good old-
fashioned paper. I'm indebted to Richard, Jenann, Philip, Frank
and Tim for allowing me to include their excellent papers.
For financial support of the London conference, I gratefully
acknowledge the generosity of the Royal Institute of Philosophy,
the British Society for the Philosophy of Science, and the LSE's
Centre for Philosophy of Natural and Social Sciences. I also thank
the London School of Economics and its Department of
Philosophy, Logic and Scientific Method for its moral and physical
support of the conference. Damian Steer also deserves thanks for
designing a beautiful conference website and helping me with all the
Preface

practical necessities on the days of the conference. For giving me a

quarter away from teaching during this time, I thank the AHRB for
a research leave award. Finally, I'm very grateful to all the
contributors to this volume and/or the conference for writing such
stimulating papers.

Craig Callender
Department of Philosophy
University of California, San Diego
La Jolla, CA 92093
USA
Notes on Contributors
Jan Faye is Senior Lecturer at the Department of Education, Philosophy
and Rhetoric, University of Copenhagen.
Huw Price is currently Professor of Logic and Metaphysics in the
Department of Philosophy at the University of Edinburgh. He also has a
continuing association with the University of Sydney, where he holds a
Personal Chair in Natural Metaphysics in the Department of Philosophy.
Frank Arntzenius is Professor of Philosophy at Rutgers University.
Tim Maudlin is Professor of Philosophy at Rutgers University.
Carol Hoefer is Senior Lecturer in Philosophy at the London School of
Economics.
Quentin Smith is Professor of Philosophy at the University of Western
Michigan.
Heather Dyke is Lecturer at the Department of Philosophy, University of
Otago.
Philip Percival is Lecturer at the Department of Philosophy, University of
Glasgow.
Nathan Oaklander is Professor of Philosophy, University of Michigan-
Flint.
Robin LePoidevin is Professor of Philosophy at The University of Leeds.
Simon Saunders is Lecturer in Philosophy of Science at the University of
Oxford. He is a Fellow of Linacre College and a Member of the
Philosophy and Physics Faculties.
Steve Savitt is Professor of Philosophy at The University of Georgia.
Yuri Balashov is Assistant Professor of Philosophy at the University of
Georgia.
Mauro Dorato is Associate Professor for the Philosophy of Science.
Richard Healey is Professor of Philosophy at the University of Arizona.
Jenann Ismael is Assistant Professor of Philosophy at the University of
Arizona.

vn
When Time Gets Off Track
JAN FAYE

Over the last forty years, philosophers have argued back and forth
about backward causation. It requires a certain structure of time for
something as backward causation to be not only possible but also to
take place in the real world. In case temporal becoming is an objec-
tive feature of the world in the sense that the future is unreal, or at
least ontologically indeterminate, it is impossible to see how backward
causation can arise. The same difficulty does not hold with respect to
forward causation. For even though it is assumed according to one
dynamic view of time, the instant view or presentism, that merely
present events exist—and past events therefore are no longer real or
have become ontologically indeterminate—such a view can still main-
tain that past events once were there to cause present events. Future
events, however, are still to come, and being indeterminate or nothing
at all, they cannot cause any events in the present. In other words,
causation backwards in time can occur only if we think of time as sta-
tic; that is, no objective becoming exists, and the world consists of
tenselessly occurring future events that exist in the same sense as past
and present events. Backward causation requires the so-called full
view, or possibly the half-full view, of time.
It has been argued that another dynamic theory of time may
allow for the logical possibility of backward causation. In his Time
and Reality (1995) Mauro Dorato says that the empty view of the
future, the position taking both past and present events to be real
and ontologically determinate in contrast to future events, may be
consistent with backward causation. It could respond to the fact that
no traces of the future have been discovered by claiming that this
empirically proves the future to be empty. Against such an argu-
ment one might still want to hold that the empty view as a dynamic
view of objective becoming does not logically open up for the exis-
tence of future events. The lack of future traces thus could not be
used as an empirical proof of objective becoming. In other words,
the reality of the future is to my mind necessary for backward cau-
sation to take place, but this does not mean that the reality of future
events would suffice for future events to be causes of present events.
The future may be real but not processes going back in time.
1
Jan Faye

I think that causality is first and foremost a common-sense notion

which has no 'natural' physical interpretation. Thus, one may
attempt to associate it with different physical notions of processes.
Four different suggestions have been put forward: (i) the causal link
can be identified with the transference of energy; (ii) it can be
identified with the conservation of physical quantities like charge,
linear, and angular momentum; (iii) it can be identified with inter-
action of forces; or (iv) it can be identified with the microscopic
notion of interaction. But it appears with respect to all four sugges-
tions that they involve descriptions that are invariant under the time
reversal operation.
No empirical evidence seriously indicates processes going back-
wards in time. This does not, however, exclude an inquiry into the
physical nature of such processes—whether they exist or not.
Today, physicists do not regard the positron-electron creation and
annihilation as an electron going back and forth in time, a sugges-
tion which Richard Feynman once forwarded as a possible interpre-
tation of the positron. The Bell type experiments have by some
been interpreted as if quantum events could be connected in such a
way that the past light cone might be accessible under non-local
interaction—not only in the sense of action at a distance but even as
backward causation. But these experiments could be interpreted in
many other ways, and as long as we do not have positive results of
post factum experiments, that is, a type of experiment where the
effect depends on something that happens later, little can be said in
favour of causal processes going backwards in time.
What is common to most of the discussions concerning possible
physical cases of backward causation is that the processes in ques-
tion are thought to be quite similar to those going forward in time.
In fact, it is nearly always taken for granted that the difference
between forward and backward causal processes has something to
do with a difference between their boundary conditions due to their
time reversal symmetry, and that the overwhelming number of
forward causal processes—with respect to backward causal
processes—depend on an unequal distribution and realization of
these boundary conditions.
Recently, Huw Price has in his Time's Arrow and Archimedes'
Point (1996) strongly advocated such a view based on the agency
theory of causation. He suggests that the causal asymmetry is an
anthropocentric matter; in nature we do not meet such an
asymmetry. Says Price: 'I shall be presenting the agency account of
causal asymmetry as a more sophisticated version of Hume's con-
ventionalist strategy. I shall argue that that it is our de facto
 When Time Gets Off Track

temporal orientation as agents that requires that we choose the

relevant conventions as we do.' (p. 158)
I shall challenge such an understanding of advanced action in
terms of a temporal symmetry of causal processes by claiming that
such a view leads to an inconsistency. I think Price is wrong; we
observe many sequences in nature which are time asymmetric and
which can be described completely non-anthropocentrically. Price
has turned the whole argument upside down: manipulation does not
conventionally define a causal asymmetry, it provides us with
knowledge of a causal orientation. He may not believe in precogni-
tion, but the fact is that we could discover it quite easily by manip-
ulation. For the sake of the argument, think of a person who is
always capable of saying in advance what will be the result of a ran-
domly selected card drawn from a deck of cards. If such a situation
really took place, I am sure that even he would agree that the later
selection of a card is the cause of the previous prediction regardless
of his conventionalist strategy. I shall, contrary to Price, argue that
the causal orientation exists objectively independent of any human
agent.
Furthermore, my claim suggests that our idea of such an objec-
tive causal direction can be identified within modern physics. I shall
propose an approach to defining retarded actions in terms of posi-
tive mass and positive energy as well as advanced actions in terms of
negative mass and negative energy—a view taking causal processes
to be asymmetric in time. According to such an alternative under-
standing, particles which could possibly manifest themselves as
going backwards in time would belong to a different category than
those which actually travel forward in time.
Not until the beginning of the twentieth century did it make
sense to imagine something as having a negative energy density or
negative mass. First the theory of relativity, and later quantum
mechanics, opened up for the theoretical possibility that mass
and/or energy could exist in two opposite forms very much as other
physical attributes such as the electric charge. Little attention,
however, has been paid to how negative mass and energy can be
understood in relation to positive mass and energy, and whether the
possible existence of negative mass and energy would have any
influence on our view on time and cosmology. Even today a basic
philosophical understanding of negative mass and energy is still
missing.
In modern physics we find two theoretical indications of negative
mass and/or energy: (i) the solutions of the four-momentum vector
do have both positive and negative values; and (ii) the zero-point
Jan Faye

energy of the quantum vacuum fluctuates: according to

Heisenberg's indeterminacy principle, the energy density of a field
must fluctuate randomly around the zero-point energy. Some
indirect effects of these fluctuations have in fact been observed,
namely the Casimir effect and the squeezed vacuum state, both
phenomena are explained in terms of negative energy.

II

What, then, is the basic challenge? I take it to be this: On the one

hand, our everyday concept of causality is such that causes are con-
sidered to be causally prior to their effects; on the other hand, the
most fundamental processes of physics are time reversal invariant.
How should we face this dichotomy? The standard interpretation
attempts to solve the dilemma by claiming that the causal direction
of reversible processes is not objective. But the consequences drawn
by the claim vary. The observed direction is said to depend on (i)
the statistics of de facto irreversible processes or on (ii) a psycholog-
ical projection of an orientation into the observation of these
processes based on our experience of human action. How can we
prove that the standard interpretation involves an inconsistency and
that it arises because of the objectivity of the causal direction?
To address this question, we must consider which conditions one
must meet in order to call the causal direction for objective. I take
the claim of objectivity to be about empirical knowledge and there-
fore its justification must be based on experience and experiments.
This claim is very much in line with Ian Hacking's pragmatic view
that atomic particles are real if we can manipulate them. Moreover,
I take the conditions of objectivity to be fulfilled if the order of the
experimental outcomes is identical with respect to different
observers having an opposite time sense. So in order to establish the
objectivity of the causal direction in time, I shall investigate the
results of a series of toy experiments.
Before we turn to these experiments, I want to clarify some
further assumptions which I think are uncontroversial in the sense
that they are acceptable even by the advocates of the standard inter-
pretation.
First, we must distinguish between process tokens and process
types. In the past much ambivalence and sloppy reasoning have
resulted from the fact that no clear distinction was made between
various descriptions of a process which is numerically the same, and
 When Time Gets Off Track

of processes which were merely qualitatively identical with one

another.
Second, we must distinguish between a passive time reversal
operation and an active time reversal operation. The difference is
that the passive transformation is one in which the same system is
described by using the opposite coordinates, whereas the active
transformation involves some physical translation, or rotation, of
the system itself.
Third, we should also, as a consequence of the above distinctions,
realize that the passive time reversal operation applies to process
tokens by referring to the numerically same process with respect to
two opposite time senses, but that the active time reversal operation
applies to process types by referring to two numerically different
tokens both belonging to the same kind.
Let us with these assumptions in mind set up the toy experiments.
Imagine two boxes, Box 1 and Box 2 facing each other, both con-
taining a shutter through which a particle can pass. At the
beginning of each run of the experiment, one of the boxes contains
a particle, but we do not know whether it is Box 1 which acts as the
emitter and Box 2 as the receiver, or vice versa. Assume, too, that the
particle is always placed in the same box—never in the other—but
that we are not told which one. It is now our job to find out whether
the particle goes from Box 1 to Box 2, or from Box 2 to Box I.
Take A to be the event that a particle is leaving or entering Box
1, i.e., A is a change taking place in Box 1, and take B to be the event
that the particle is leaving or entering Box 2, i.e., B is a change tak-
ing place in Box 2. If no such activity occurs in the respective boxes_
irrespectively of whether they are open or closed, we shall call it A
and B. We can find out whether a particle has left or entered a box
by weighing the box because of the loss or gain of the particle's
energy E, therefore whether A or B corresponds with a decrease or
an increase of energy.
Do we have some common sense criteria which reveal in which
direction the particles are travelling? If so, are these criteria objec-
tive in the sense that even an observer of the opposite time sense can
rely on the same criteria? The answer seems to be yes. We would say
that A is the cause of B whenever:
(i) A happens even if Box 2 is closed,
(ii) B_ does not happen unless Box 1 is open,
(iii) A happens if and only if Box 1 is closed,
(iv) B may happen even if Box 2 is open.
Jan Faye

I hold that if our observations meet these criteria we have acquired

knowledge of a causal direction between As and Bs.
Now, consider the assumption that the particles move from Box 1
to Box 2. To test this hypothesis we must run a series of four exper-
iments in which we vary between opening and closing the shutters.
By doing this we get four different combinations: (1) closed, closed;
(2) open, open, (3) closed, open; and (4) open, closed. As shown in
figure 1 we wouldexperience four sets of events in these four cases
(A,B);(A,B);(A,B);and(A,B);

BOX 1 BOX 2 BOX1 BOX 2

AE=0 AE=0 A

CO S3 |
AE<0 AE>0 A

AE=0 AE=0 A
CO | 03 I

AE<0 AE=0 A

Figure 1. A causes B (The normal oberver).

From this we can conclude that particles do in fact move from Box
1 to Box 2. The reason is that we have two As but only one B, and
that one B appears even though Box 2 is open and Box 1 is closed;
whereas one A appears when Box 1 is open and Box 2 is closed. In
my opinion it is a series of experiments like these which justify our
belief in the particles moving from Box 1 to Box 2.
In addition, we may notice that the cause A seems to coincide
with a decrease of energy while the effect JB seems to coincide with
an increase of energy. Perhaps we could identify causation in
physics with propagation of energy? We shall soon return to this
point.

Ill

In the mean time we must find out whether our beliefs of the causal
direction is about objective matters or not. We shall investigate how
an observer with the opposite time sense of ours (a counter observer)
 When Time Gets Off Track

would describe the situation. As I said, there are two very different
ways of understanding time reversal invariant, and I shall argue
that it is only the passive operation which corresponds with a descrip-
tion according to the observer with the opposite view of time.
One way of understanding time reversal invariant is as an active
transformation of the original process under consideration so the
particles now move from Box 2 to Box 1. The active time reversal
operation is illustrated in figure 2.
BOX 1 BOX 2 BOX 1 BOX 2

AE=0 AE=0

AE>0 AE<0

AE=0 AE<0

AE=0 AE=0

Figure 2. B causes A (The normal observer)

Here the situation is such that B is now the cause of A based on

exactly the same reasoning as before, and B can also be, or so it
seems, identified with a loss of energy and A with a gain of energy.
The problem for the standard interpretation is that this description
does not give us the opposite description of the same tokens accord-
ing to a different time sense. It presents rather a description of dif-
ferent tokens with respect to the same time sense as ours.
Thus, the second way of understanding time reversal invariant is
as a passive transformation of the original process in question. So
what gives us the description of the opposite time sense is the pas-
sive time reversal operation with respect to the same tokens. Figure
3 shows how the standard interpretation will describe the situation.
The counter observer with the opposite time sense will see B as a
loss of energy and A as the gain of energy. Hence it seems as if Box
1 is now acting as the receiver and Box 2 as the emitter. This is sim-
ilar to the situation where we noticed the active time reversal oper-
ation described (apart from the fact that Box 1 will now receive a
particle, not from Box 2, but from the environment or the infinity).
Thus the standard interpretation draws the following conclusion
Jan Faye

BOX 1 BOX 2 BOX1 BOX2

AE=0 AE=0 A B

AE>0 AE<0 A B

AE=0. AE=0 A B

AE>0 AE=0 A B

Figure 3. B causes A (The counter observer)

about the counter observer: he will observe particles move from Box
2 to Box 1 according to the laws of nature but going forward in his
time. In other words, the counter observer will describe the causal
direction of the particles opposite to what a normal observer would
do. The conclusion is therefore that the causal direction of the very
same tokens cannot be objective.
I believe, nonetheless, that this analysis is flawed. The experi-
ment provides the counter observer with no criteria for making such
a judgement. He will face a situation where Box 2 is open but no
changes take place in this box unless Box 1 is open; whereas Box 1
may be open and still undergo a change without Box 2 being open.
How then can the counter observer claim that Box 2 is the emitter?
He cannot! In my opinion even the counter observer must admit
that the causal direction is objective because he will ascribe the
process with the same orientation as the normal observer. He will,
like the normal observer, observe two As in Box 1 and only one B in
Box 2. Furthermore, he will realize that even though Box 2 is open,
it is not always the case that a change happens (no loss of energy) in
that box. A change happens in Box 2 only if Box 1 is open; but in
Box 1, if it is open, a change happens regardless of whether or not
a change happens in Box 2; that is, regardless of whether Box 2 is
open or not. But this combination of events formed the exact criteria
that the normal observer used to reason that the particles were mov-
ing from Box 1 to Box 2. So the counter observer will see the same
experimental pattern as the normal observer, and, by using the same
criteria, reach the same conclusion as the normal observer about the
direction of the process. We can therefore conclude that the parti-
cles move objectively from Box 1 to Box 2. According to the counter
 When Time Gets Off Track

observer, however, the propagation is oriented backwards in his

time as opposed to forward in time for a normal observer.
Apart from the temporal orientation, the normal observer and the
counter observer will also differ with respect to the ascription of
physical properties to the particles. We should be aware of the fact
that the theory of relativity yields two solutions of the four-momen-
tum vector with respect to energy: (i) positive energy solutions, and
(ii) negative energy solutions. As a consequence, Feynman's propaga-
tor formalism usually integrates over all positive energies into the
future and over all negative energies into the past. These different
solutions can easily be understood as the time sense reversed descrip-
tion of one and the same particle. Feynman himself suggested, based
on this formalism, that a positron could be interpreted as a negative
energy electron going backwards in time. As the above thought exper-
iment shows this is a false idea. Nothing of relevance is changed in
the experiment if one replaces the particles with positrons or any
other kind of anti-particles. Instead we should think of the positive
energy solutions as how the normal observer will describe a retarded
particle (the usual common particles or their anti-particles) and of the
negative energy solutions as how the counter observer will describe
such a retarded particle. (As illustrated in figure 4 and 5.)
BOX 1 BOX 2 BOX1 BOX 2

AE*=0 AE *=o A B

AE' <C AE *>0 A B

AE*=0 AE *=0 A B

AE* <c AE *=0 A B

Figure 4. A causes B (The normal observer)

The objective interpretation connects energy and the time

reversal operations in the following way:
(a) The passive time reversal operation turns a positive energy
particle with positive mass into a negative energy particle with
negative mass, or vice versa, by viewing the same process
token from the opposite frame of time sense.
Jan Faye

BOX 1 BOX 2 BOX 1 BOX 2

AE-=0 AE*=0 A B

AE">C AE"<0 A B

AE"=0 AE-=0 A B

AE">C AE"=0 A B

Figure 5. A causes B (The counter observer)

(b) The active time reversal operation turns a positive energy par-
ticle with positive mass into a positive energy particle with pos-
itive mass (or a negative into a negative) by providing a
description of another process token seen from the same frame
of time sense in virtue of reversing the particle's three
momentum vector and spin vector.
These two operations also explain how the normal observer and the
counter observer can—based on their different perspectives—agree
that the causal direction is objective. Both can identify the causal
direction of a process with the direction of its positive energy. So a
normal observer will see a retarded particle going forward in her
time along its positive energy component, whereas the counter
observer with the opposite time sense will regard the same particle
as going backwards in his time. For even though the counter observ-
er will 'see' the particle along its negative energy component for-
ward in his time, the direction of causation coincides with the posi-
tive energies backwards in his time—the direction which the com-
mon criteria point out is the causal direction.

IV
I want to make some further comments on the interpretation of the
above experiment and on backward causation in general.
First of all, the sceptic may, as Lars-Goran Johansson has pointed
out to me, raise the following objection: it seems to me that your
10
 When Time Gets Off Track

criteria rest on a false assumption which invalidates your conclu-

sion. Think of a photon instead of a particle with a rest mass.
According to the toy experiment a photon may leave Box 1 even if
it is not absorbed by Box 2. That means that there must be some
environment functioning as absorber. This environment can just as
well emit a photon. Hence, a photon may very well be absorbed in
Box 2 even if Box 1 is closed; that is, we may have an event B with-
out an A. So the above conditions (i) and (ii) are incompatible in the
settings of current physical theory. But this is devastating for your
argument, since the asymmetry between cause and effect in your
model now build upon the assumption that the effect, viz. the
absorption of energy (in the normal case), may occur even though
the 'cause', viz. the emission of energy from Box 1, has not
occurred, whereas the cause, the emission of energy, cannot occur
without the effect. If this asymmetry is dismissed, you have no
ground, in your model, to tell what is cause and what is effect, inde-
pendently of the time ordering.
Now, the sceptic may continue his criticism by saying: you could,
indeed, add as an explicit criterion for the entire experiment that the
environment is a perfect black body; however, would such a response
not undermine your argument, too? For there are not real perfect
black bodies, and so this model cannot even be used as a thought
experiment showing the possibility for an objective time ordering.
How can this sceptical challenge be met? One should be aware of
a couple of things. The sceptic is not presenting a case based on the
original tokens. The case where Box 1 is closed and Box 2 is open
but receives a photon from the environment is different from the
original case in which Box 1 was closed and Box 2 open but did not
receive a particle. We can also in the latter case find out whether Box
2 is absorbing or emitting a photon by looking for the gain or the
lost of weight. So what the sceptical objection amounts to is a
request for an explanation of how the following series of event
tokens can be excluded: (A, B); (A, B); {A, B) and {A, B). For if
the toy experiment produces such a series of events, we will not be
able to use the above criteria to find the causal order.
A solution is, I think, within our range. Although it is possible
that Box 1 may be closed and Box 2 is open but nevertheless receive
a photon, it may not be as likely a situation as the case where Box 2
receives no photon. We can solve the problem by imagining (a com-
bination of) three possibilities:

1. We may think of the environment as a black-body. But, again,

there are no perfect black-body absorbers.
11
Jan Faye

2. We can argue that the experiment has been designed so that

probability of Box 2 containing a photon depends on whether
Box 1 is open or closed. With respect to Box 2 we have: P(pho-
ton in box 21 box 1 is open) » P(photon in box 21 box 1 is
closed), whereas with respect to Box 1 we have: P(photon in
box 11 box 2 is open) = P(photon in box 11 box 2 is closed). We
have not thereby excluded that the environment could emit a
photon which appears in Box 2 even though Box 1 is closed.
3. Run the experiment within four time intervals ATn , n = 1,2,3,
4, during which one or two boxes will be open (if they are
going to be open) and be weighed. The length of the intervals
will be around 2 l/c, long enough for a photon to travel back
and forth between the two boxes, but short enough for any
photon from the environment to play an active role in the
experiment, that is, its probability would be very low.
The sceptic will most certainly reply that such a response misses the
point. The core idea of the proposed refinement is to claim that
probabilities for emission and absorption depend on earlier events
but not on later events. This is problematic when talking about pho-
tons, i.e., electromagnetic quanta. Let us look at the world from the
point of view of the photon for a while. As it travels with the speed
of light, all distances are reduced to zero which means that in a
sense a photon is present everywhere at each moment of time. That
is another way of saying that the environment must be taken into
account as absorber or emitter v/hen only one box is open. In this
context an important feature of quantum field theory is that the
probability for excitation of an atom at one point starts increasing
at the very same moment a quantum is emitted from another atom
at another place, independently of their distance. (See G.
Hegerfeldt (1994)). The fatal consequence for my argument being
applied to photons seems to be that the situation becomes symmet-
ric in excitation/deexcitation so that the probability for a photon
being present in Box 1 is not independent of the opening/non-open-
ing of Box 2. In short, one runs into trouble by assuming a particle
view of photons.
But, as we have seen, it is not really necessary to use photons. We
have considered particles with rest mass and my argument certainly
goes through. This raises, however, an interesting problem. For if
we regard photons as particles, we arrive at a very different conclu-
sion about the asymmetry of emission and absorption than we do
when taking particles with rest mass into account. What does this
mean? Is my argument and the common understanding of quantum

12
 When Time Gets Off Track

field theory consistent with one another? This is still something to

consider.
I think there is no problem here. It might very well be that the
environment can act as emitter of photons and therefore that one
photon may occur in Box 2 even though Box 1 is closed. But one has
to remember that by observing the thought experiment both the
normal observer and the counter observer see the same token
processes which means that if the counter observer watches that a
change takes place in Box 2, in spite of the fact that Box 1 is closed,
so will the normal observer. An unwanted situation like this may be
avoided by a long series of experiments or by increasing the num-
ber of the photons to work with a beam instead of single photons at
a time. It seems after all possible to establish the suggested asym-
metry of emission and absorption between Box 1 and Box 2.
A second objection may be raised against my notion of an active
transformation by not accepting the expression: 'the active transfor-
mation involves some physical translation or rotation of the system
itself.' Here I seem to speak of translations and rotations in physi-
cal space, but what has that to do with time reversals?
The passive time transformation is 'one in which the same system
is described by using the opposite coordinates.' This applies to the
time reversal operator utilized in quantum mechanics. My interpre-
tation of this operation seems to rest on a confusion of two things:
the change of time coordinates is a change of representation of the
time relations 'before' and 'after', not an exchange of before and
after. Normally we represent the times of events in such a way that
if the event a occurs before event b, the time of a is represented by
a smaller number than the time of b. Time reversal is reversal of
this convention: later times are represented by smaller numbers,
(the time relation ... earlier than ... is represented by the numerical
relation ...>...) It is not a reversal of the real ordering before-after,
or of the perception of the real ordering of events. It could not pos-
sibly be because in a mathematical theory one must represent every-
thing 'real' with mathematical entities such as numbers and func-
tions and no operation on these entities can by itself, i.e., without
further interpretative assumptions, be taken to represent something
extra-mathematical such as a real change of time direction.
I do not believe, however, that I conflate two things when talking
about time reversal: It has been argued by many philosophers and
scientists that the time reversal of mechanical processes should be
understood as if the planets suddenly changed their momentum in
the opposite direction around the sun. This is what I mean by the
active transformation. I don't think that the example says anything
13
Jan Faye

about time reversal. It represents, as I pointed out long ago, a case

of motion reversal by involving two different tokens of the same
process type and the illustration with the planets rests on the ambi-
guity between types and tokens.
When it comes to the passive transformation, it is correct to dis-
tinguish the difference between the exchange of representation of a
system and what can be called the exchange of before and after. In
fact, this is what I am trying to argue when I claim that the causal
direction of the process token is objective with respect to observers
with opposite time. We have therefore two questions to answer: (i)
Is the reversal of time merely a subjective convention? And if it
isn't, what then does it represent?
What I want to say is this: One may exchange the temporal rep-
resentation of the system by changing the mathematical representa-
tion of the time relations 'before' and 'after'. Nonetheless, a
philosopher like Price takes this to mean that the physical laws
themselves are time symmetric. It is only the distribution of bound-
ary conditions which makes a temporal asymmetry between before
and after.
Contrary to this view, I believe that the opposite mathematical
representations should be interpreted as if they were descriptions
made by observers moving opposite one another in time. If one
wants the representation to reflect how observers with opposite time
senses will represent the time reversed situation, one also needs to
change the sign of energy as a reflection of their different observa-
tions. But this is not a real change of before and after because, in my
vocabulary, the causal direction is objective by being constituted by
the positive energy component.
Some people may have difficulties in understanding what is
meant by saying that time goes backwards, entirely or for a particu-
lar person. Would a person with a reversed sense of time die before
he is born? What is then meant by the terms 'dying' and 'being
born' ? Would such a person remember what will happen tomorrow
but not what he had for breakfast the very same morning? And
would he say that he remembers the future, or would he say: 'I, as
everybody else, can only remember what has passed. It is only that
what I call the past, you call the future?' Surely, the issue is not
about what words we use!
I think that some of these questions reflect real difficulties in our
understanding of what a person with a differing time sense means:
but I also think that they are based on our common ways of seeing
things. A counter observer will, of course, see his life as we experi-
ence ours. He will see himself grow older and eventually die. But a
14
 When Time Gets Off Track

normal observer (as we are) will certainly 'observe' such a person

becoming younger and eventually being born.
One may also question the proposal of seeing causation in physics
as something that has to do with energy transference. Against the
idea of identifying the causal direction of a physical system with the
flow of positive energy through time, I have been confronted with
two objections that I shall address here.
Phil Dowe has argued that in a static world in which a physical
system S contains the same amount of positive energy at t\ and t2—
say, S is a single particle which is represented by a straight world-
line—no information about the energy can tell us whether t^ or t2 is
later than the other, and ipso facto it makes no sense to talk about the
flow of energy from ^ to t2. I think the point is well taken.
Apparently to say that the same amount of energy in the system S
is propagating from t\ to t2 also requires a dynamic flow of becom-
ing. I even think that the objection can be phrased in more general
terms. As long as we describe the energy from the internal perspec-
tive of S itself, and no redistribution of energy takes place within S,
we cannot point to what may determine the transference of energy
from t-[ to t2, because the fact that the energy is the same at two dif-
ferent moments does not involve any direction.
I take, however, the right response to be this: the positive energy
merely endures between t\ and t2, hence no causal changes take
place during this interval. From its own perspective 5 does not
experience any lapse of time, t\ will be identical with t2, and there-
fore it will not causally move between ^ and t2. Thus, a physical
system—or the entire world for that matter—described from its
own internal perspective exists timelessly unless its energy is redis-
tributed within itself. But from an external perspective of S, as soon
as S occupies two different places, Xj and x2, at t\ and t2 we have a
transference of energy. Assume that we find that at tx the energy is
positive in xx and zero in x2, and that at t2 the situation is reversed,
then we know that a transference of positive energy took place
between t^ and t2, and by using the above experiments we can get to
know in which direction this flow happens; that is, from f j to t2, or
vice versa.
Another objection is due to Huw Price. One may, he says, refuse
to accept the above interpretation by arguing that the sign of energy
merely builds on a convention. How can the direction of causation
be identical with the direction of positive energy in case we can
replace a positive description with a negative one? We merely call
positive energy negative and negative energy positive. Indeed we
might. But this linguistic fact does not suffice to refute the
15
Jan Faye

interpretation suggested. It can be compared to the choice we can

make between various geometries for the description of the physical
space. Such a selection is also based on a convention. But when the
convention first has been established, we must stick to the choice we
once made. Let me remind you of Carnap's remarks about the con-
ventionalist view concerning the choice of geometries. As he points
out: 'Because Poincare said the choice was a matter of convention,
his view became known as the conventionalist view. In my opinion,
Poincare meant that the choice was made by the physicist before he
decided which method to use for measuring length. After making the
choice, he would then adjust his method of measurement so that it
would lead to the type of geometry he had chosen. Once a method of
measurement is accepted, the question of the structure of space
becomes an empirical question, to be settled by observation.' (p. 160)
This is precisely to say what the identification of the causal direction
with the propagation of positive energy is all about.
Finally, we are now in a position to grasp how a notion of back-
ward causation should be accommodated into physics. Since posi-
tive mass and positive energy states forward in time can be identi-
fied with the causal direction of processes forward in time, it means
that positive mass and positive energy states backwards in time can
be associated with processes propagating backwards in time. A
counter observer will, according to his time sense, see such process-
es of advanced particles moving forward in his time. If a normal
observer were going to see such processes of advanced particles,
they would appear to her as having negative masses and as a series
of negative energy states succeeding one another forward in her
time. In spite of this the normal observer is forced by the choice of
convention to admit that the advanced particles really go backwards
in her time through their positive energy states.
In my opinion whether backward causal processes exist or not is
an objective question which in principle can be settled by
experiments. It is neither a matter of convention, nor is it a matter
of subjective projections, or even a matter of the violation of de
facto irreversibility like entropy and expanding outgoing waves. The
existence of backward causation will prove itself, I suggest, when-
ever we observe a violation of the conservation of energy. Only then
do we see time gets off track.

Acknowledgement: I wish to thank Mauro Dorato and Lars-Goran

Johansson for their critical comments and suggestions to an earlier
version of this paper.
16
 When Time Gets Off Track

References
Carnap, Rudolf 1966. Philosophical Foundations of Physics. New York,
London: Basic Books.
Dorato, Mauro 1995. Time and Reality, Bologna: Clueb.
1998. 'Becoming and the Arrow of Causation' (unpubl.)
Faye, Jan 1989. The reality of the future. Odense: Odense University Press.
1997a. 'Is the Mark Method Time Dependent?', in Jan Faye, Uwe
Scheffler & Max Urchs (eds) Perspectives on Time. Boston Studies in the
Philosophy of Science, 189, Dordrecht: Kluwer, 215-36.
1997b. 'Causation, Reversibility and the Direction of Time', in Jan
Faye, Uwe Scheffler and Max Urchs (eds) Perspectives on Time, 237-66.
Hegerfeldt, G. 1994a. 'Causality Problems for Fermi's Two-atom System',
Phys. Rev. lett. 72, no. 5, 596-99.
1994b. 'Remark on Causality and Particle Localization.' Phys. Rev. D,
10, no. 10, 3320-21.
Price, Huw 1996. Time's Arrow and Archimedes' Point. Oxford: Oxford
University Press.

17
Burbury's Last Case: The Mystery of
the Entropic Arrow
HUW P R I C E

Does not the theory of a general tendency of entropy to diminish

[sic1] take too much for granted? To a certain extent it is sup-
ported by experimental evidence. We must accept such evidence
as far as it goes and no further. We have no right to supplement it
by a large draft of the scientific imagination. (Burbury 1904, 49)

1. Introduction
Samuel Hawksley Burbury (1831-1911) was an English barrister
and mathematician, who favoured the latter profession as loss of
hearing increasingly curtailed the former. The Bar's loss was
Science's gain, for Burbury played a significant and perhaps still
under-rated part in discussions in the 1890s and 1900s about the
nature and origins of the Second Law of Thermodynamics. One
commentator of the time, reviewing Burbury's The Kinetic Theory
of Gases for Science in 1899, describes his role in these terms:
[I]n that very interesting discussion of the Kinetic Theory which
was begun at the Oxford meeting of the British Association in
1894 and continued for months afterwards in Nature, Mr.
Burbury took a conspicuous part, appearing as the expounder
and defender of Boltzmann's H-theorem in answer to the ques-
tion which so many [had] asked in secret, and which Mr
Culverwell asked in print, 'What is the H-theorem and what does it
prove?' Thanks to this discussion, and to the more recent publi-
cation of Boltzmann's Vorlesungen iiber Gas-theorie, and finally to
this treatise by Burbury, the question is not so difficult to answer
as it was a few years ago. (Hall 1899, 685)
In my view, however, it is at best half-right to describe Burbury as
a defender of the H-Theorem. In some respects, he was the leading
advocate for the prosecution. The crucial issue arising from
Culverwell's (1890a, 1890b, 1894) enquiry was the source of the
1
By 'entropy' here, Burbury seems to mean the quantity H of H-
Theorem fame, which diminishes as a gas approaches equilibrium.
19
Huw Price

time-asymmetry of the H-Theorem, and while Burbury put his

finger on the argument's time-asymmetric premise, he himself
fingered it—in true forensic spirit—as an object of considerable
suspicion. A decade later, as the quotation with which we began
indicates, he still wasn't convinced that we are entitled to assume it;
and in their important survey article of 1912, Ehrenfest and
Ehrenfest note his continuing dissent on the matter (1959, 42; 95, n.
168). Indeed, in reading Burbury's work from the late 1890s and
1900s, one gets the sense that the issue of the basis and justification
of the Second Law remained the great intellectual puzzle of his later
life.
Later in the paper I want to return to Burbury's contribution to
the debate of the 1890s. I want to endorse his scepticism about our
present entitlement to assume that entropy will continue to increase,
and I want to show that in one important respect, his contribution to
the debate has been systematically misunderstood. (In uncovering
this misunderstanding, I shall argue, we undermine one widespread
conception of what it would take to account for the Second Law.)
However, the main task of the paper is to call attention to two
kinds of objection to some well-known strategies for explaining the
temporal asymmetry of thermodynamics. Neither of these objec-
tions is particularly novel, but I think that despite the long history
of the problem, neither has received the prominence it deserves. As
a result, in my view, the strategies concerned continue to enjoy more
popularity than they merit.
In a more constructive vein, I want to locate the strategies I am
criticizing within a kind of taxonomy of possible approaches to the
problem of the thermodynamic asymmetry. In this way, I want to
suggest that they are mistaken not only for the reasons identified by
the two objections in question, but also for a more basic reason. In
effect, they are trying to answer the wrong question. This may sound
doubly critical, but the double negative yields a positive—given that
the proposed strategies don't work, it turns out to be good news that
we don't need them to work, in order to understand the thermody-
namic asymmetry. In this respect, the present paper deals with a
topic I have examined in more detail in other recent work (Price
2002): the issue as to the precise nature of the time-asymmetric
explanandum presented to us by thermodynamic phenomenon.

1.1 Origins of the problem

The time-asymmetry of thermodynamics is associated with the
Second Law. According to this principle, non-equilibrium systems
progress monotonically towards equilibrium. Temperature differ-
20
 Burbury's Last Case: The Mystery of the Entropic Arrow

ences decrease, energy concentrations dissipate, and entropy

increases monotonically until equilibrium is reached.
In the latter half of the nineteenth century, having recognized
and described this tendency as a phenomenological principle,
physics sought to explain it in statistical mechanical terms. The
Second Law turned out to be different from more familiar laws of
physics in at least two ways. First, it was probabilistic rather than
strictly universal in nature—exceptions were possible, though very
unlikely. Secondly, and more interestingly for our present purposes,
it was time-asymmetric—the phenomena described by the Second
Law showed a clear temporal preference. Late in the nineteenth
century, and especially in the debate mentioned above, physics
began to see that the latter fact is rather puzzling, in the light of the
apparent time-symmetry of the laws of mechanics. How could
symmetric underlying laws give rise to such a strikingly time-
asymmetric range of phenomena as those described by the Second
Law?
More than a century later, there is surprisingly little consensus as
to how this question should be answered. Late in twentieth century,
indeed, a leading authority on the conceptual foundations of
statistical mechanics could still refer to the puzzle of the time-asym-
metry of the Second Law as the elusive object of desire. (Sklar
1995) In my view, the fact that a solution has remained elusive rests
in part on some confusion about what we are actually looking for. It
is unusually difficult to be clear about what precisely needs to be
explained about the time-asymmetry of the Second Law. There are
several competing conceptions of what the problem is, with the
result that proponents of rival approaches tend unwittingly to be
talking at cross-purposes.
In order to try to clarify matters, I proceed as follows. I focus first
on the nature of the time-asymmetric phenomena that are the
source of the problem. Here, I think, a few simple remarks do a
great deal to help us keep the true object of desire in view, and to
avoid issues which are not directly relevant to the puzzle of time-
asymmetry. I then distinguish two major competing approaches to
the explanation of these time-asymmetric phenomena—two
approaches which differ markedly in their conception of what needs
to be done to solve the puzzle.
Despite their differences, the two approaches do agree about one
part of the puzzle—they both hold that an important contributing
factor to the observed thermodynamic time-asymmetry is that
entropy was much lower in the past than it is now. However, where-
as one approach argues that this asymmetric boundary condition is
21
Huw Price

the sole time-asymmetric source of the observed asymmetry of

thermodynamic phenomena, the other approach is committed to the
existence of a second time-asymmetry—a time-asymmetric lawlike
generalization. So far as I know, the distinction between these two
approaches has not been drawn explicitly by other writers. Without
it, it is not easy to appreciate the possibility that many familiar
attempts to explain the time-asymmetry of thermodynamics might
be not mistaken so much as misconceived—addressed to the wrong
problem, in looking for time-asymmetry in the wrong place.
In my view, the 'two-asymmetry' approaches are misdirected in
just this way. In looking for a basis for a time-asymmetric general-
ization, they are looking for something the explanation of the phe-
nomena in question neither needs nor wants. However, my main
aim is not to argue that these approaches address the problem of the
thermodynamic asymmetry in the wrong terms, but to show that
they don't succeed in addressing it in their own terms—they fail by
their own lights. Specifically, I shall raise two objections to the two-
asymmetry approach. Neither objection applies to all versions of
this approach. However, I shall note two characteristics, one or
other of which seems a feature of any version of the two-asymme-
try view. Each characteristic leaves the view in question open to one
of my two objections. My claim is thus that any version of the two-
asymmetry approach is subject to at least one of the two objections.
Some, as we shall see, are subject to both.
As I said, however, I take these negative conclusions to be good
news for the project of trying to understand the thermodynamic
asymmetry itself. For they bolster the case—already strong, in my
view, on simplicity grounds—for taking the one-asymmetry view to
provide the better conception of what actually needs to be
explained.

2. Three Preliminary Clarifications

I begin with three preliminary remarks about the nature of

thermodynamic time asymmetry. The purpose of these remarks is
to 'frame' the relevant discussion, as I see it, and explicitly to set
aside some issues I take not to be of immediate relevance. I am
aware, of course, that these judgements of relevance themselves are
not uncontroversial, and that some readers will feel that I am ignor-
ing the interesting part of the subject. But in territory thick with the
criss-crossed tracks of previous expeditions, it is useful to all parties
to mark out one's own course as clearly as possible. What follows are
22
 Burbury's Last Case: The Mystery of the Entropic Arrow

three assumptions that I shall take for granted in the remainder of

the paper.

2.1 Numerical asymmetry not practical reversibility

It is common to characterize the time-asymmetry of thermody-
namic phenomena in terms of the irreversibility of the processes by
which matter tends to thermodynamic equilibrium. However, the
term 'irreversible' is ambiguous. In particular, some writers inter-
pret it in a very practical sense. The time-asymmetry of thermody-
namics is hence thought to be tied to the practical difficulty of
'reversing the motions' in real systems. (Ridderbos and Redhead
1998 appear to take this view, for example, focusing on the contrast
between most ordinary systems and those involved in the so-called
spin-echo experiments, which do permit reversal of motions, at
least to an unusual degree.)
In my view, this focus on practical reversibility mislocates the
important time-asymmetry of thermodynamic phenomena.
Consider a parity analogy. In a world containing handed structures
of a certain kind, we may distinguish two sorts of question: (i) Can
a left-handed example of such a structure be 'reversed' into a right-
handed version, and vice versa? (ii) Is there a numerical imbalance
in nature between the number of left-handed and right-handed
examples? These two questions are logically independent—one can
easily imagine worlds with any of the four possible combinations of
answers.
Similarly in the temporal case, I think. The issue of the practical
reversibility of a time-oriented phenomenon is logically indepen-
dent of that of the numerical imbalance in nature between examples
of the phenomenon in question with the two possible temporal ori-
entations. Clearly, our world shows a vast numerical imbalance of
this kind, for the kind of phenomena described by the Second Law.
It is this numerical imbalance that is the primary puzzle, in my
view, not the issue of the practical reversibility of individual sys-
tems.

2.2 A monotonic entropy gradient, not an increase or decrease

When we say that entropy always increases, or that entropy change is
always non-negative, we presuppose a convention as to which is to
count as the 'positive' direction on the temporal axis. If we reverse
this convention, an increase is redescribed as a decrease, a non-neg-
ative change as a non-positive change. In one rather uninteresting
23
Huw Price

sense, this means that the Second Law is itself a conventional matter—
in its standard formulations, it depends on a convention concerning
the labelling of the temporal axis.
Some people may feel that they can make sense of the view that
the choice of labelling is not merely conventional—that one or other
labelling is objectively correct, and that time itself is objectively
'directed' in this way. The point I wish to emphasize is that these
are separate issues from that of the thermodynamic asymmetry. In
particular, the objectivity of the thermodynamic asymmetry does
not depend on the view that time itself is objectively directed (what-
ever that might mean).
Indeed, the thermodynamic asymmetry is easily characterized in
a way which avoids the conventional choice of labelling altogether.
We simply need to say that the entropy gradients of non-equilibri-
um systems are all aligned in the same direction, leaving it unspec-
ified (and a conventional matter) whether they all increase or all
decrease. (By way of analogy, we might describe a universe contain-
ing nothing other than a single hand as objectively 'handed', while
regarding it as a conventional matter whether it contains a left hand
or a right hand.)

2.3 'Entropy' is inessential

One source of complexity in discussions of the Second Law is the
existence a variety of competing definitions of entropy. In view of
this complexity, it is helpful to keep in mind that if necessary, the
puzzling time-asymmetry of thermodynamic phenomena can be
characterized without using the notion of entropy—at least if we are
prepared to tolerate some loss of generality. We can describe the
puzzle by being more specific—by listing some of the actual kinds
of physical phenomena that exhibit a temporal preference; that
occur in nature with one temporal orientation, but not the other.
Warm objects cool down rather than heat up in a tub of ice, pres-
surized gas flows out from but not into a bottle, so on. (In all these
cases the description presupposes the ordinary temporal labelling,
in the way noted in Section 2.1.)
The notion of entropy may turn out to provide a useful way of
generalizing over this class of phenomena, but doing without it
wouldn't deprive us of a way of talking about the temporal bias dis-
played by real systems—which, after all, is the real source of the
puzzle. I shall continue to use the term in this paper, but take it to
be merely a place-holder for more specific descriptions of the rele-
vant properties of time-asymmetric thermodynamic systems.
24
 Burbury's Last Case: The Mystery of the Entropic Arrow

3. Two Models of the Origin of the Thermodynamic

Asymmetry

In this section I distinguish two different explanatory models for

the thermodynamic asymmetry. In my view, insufficient attention
to this distinction does a great deal to explain the striking lack of
consensus about this topic—in particular, it explains the major
respect in which various participants in the debate have been talk-
ing at cross purposes.

3.1 The two-asymmetry view

As noted above, the nineteenth century attempt to reduce thermo-
dynamics to statistical mechanics led to two realizations about the
Second Law: first, that it is probabilistic rather than strictly
universal in nature; and second, that it is time-asymmetric. As a
result, the Second Law is commonly seen as a time-asymmetric
probabilistic generalization, with a lawlike or quasi-lawlike charac-
ter—a general constraint on the behaviour of matter, preventing or
at least discouraging (in a probabilistic sense) entropy from decreas-
ing, in the ordinary time sense.
What does 'lawlike' mean in this context? This question deserves
more attention than I can give it here, but for present purposes the
crucial feature seems to be projectibility. At the very least, for two-
asymmetry views, the Second Law retains the status of a time-
asymmetric generalization on which we are entitled to rely, in forming
expectations about the future behaviour of matter. (As we shall see,
it cannot have this status for one-asymmetry views.)
The nomological character of two-asymmetry approaches some-
times also shows up as the view that matter has a certain time-asym-
metric property, or disposition—that of being 'thermodynamic'
rather than 'anti-thermodynamic'. Thus Richard Feynman, in com-
ments attributed to 'Mr X' in discussion reproduced in (Gold
1963), refers to 'the assumption that matter is thermodynamically
"one-sided", in the ordinary sense that it damps when you try to
shake it.' (Gold 1963, 17) Here it is the property which is thought of
as projectible—the time-asymmetric disposition to absorb heat,
disperse energy, and the like.
According to this picture, then, a large part of the task of explain-
ing the observed thermodynamic asymmetry is a matter of finding
a basis for a time-asymmetric nomological generalization or dispo-
sition—finding something that 'makes' entropy increase towards
the future, which 'makes' matter thermodynamic, in the above
25
Huw Price

sense. Whatever this basis is, it needs to be time-asymmetric itself.

Otherwise, the generalization would hold in both temporal direc-
tions, and entropy could only be constant. (Damping becomes anti-
damping under time reversal, so matter that damped in both time
directions would be incompatible with the existence of shakers.)
However, it is crucial to note that even a time-asymmetric nomo-
logical constraint of this kind would not give rise to the observed
thermodynamic asymmetry unless entropy were initially lower than
its maximum possible value—unless there were something around
to do the shaking, in Feynman's example. Otherwise, the effect of
the constraint would simply be to maintain a state of equilibrium,
with no observable entropy gradient, and no asymmetry. So in order
to explain the observed time-asymmetry, according to this concep-
tion of the Second Law, we actually need two time-asymmetric
ingredients: the asymmetry nomological tendency or generalization,
and an asymmetric boundary condition, to the effect that entropy is
low at some point in the past (again, in the ordinary time sense).
Diagrammatically, the observed time-asymmetry thus arises as
follows, according to the two-asymmetry view:

Asymmetric boundary condition—entropy low in the past

+ Asymmetric lawlike tendency—entropy constrained to increase

Observed asymmetry.

As we shall see, these two-asymmetry views share the need for the
first component—the asymmetric boundary condition—with their
'one-asymmetry' rivals. Their distinguishing feature is thus the
attempt to find a basis for the second component—for a nomologi-
cal time-asymmetry. (In practice, this is often seen as very much the
more important aspect of the problem.)
In my view, the search for such a time-asymmetric principle or
property is both misconceived and unsuccessful. It is unsuccessful
in the light of one or both of the two objections I want to raise
below to different versions of this two-asymmetry approach to the
explanation of the thermodynamic asymmetry. And it is miscon-
ceived because there is a preferable alternative view of the origins of
the observed time-asymmetry of thermodynamic phenomena,
which—because it does not involve a time-asymmetric generaliza-
tion in the first place—simply does not face the problem of a basis
or explanation for such a generalization.2
2
In Price 2002 I call the two-asymmetry view that the 'causal-general'
approach, and the one-asymmetry view the 'acausal-particular' approach.
26
 Burbury's Last Case: The Mystery of the Entropic Arrow

3.2 The one-asymmetry view

We noted that a nomological time-asymmetry is at best half of what
we need to explain the observed asymmetry of thermodynamic phe-
nomena. A constraint preventing entropy from decreasing does not
yield an asymmetry unless entropy is low to start with. This low
entropy past boundary condition is the second of the two-asymme-
try approach's two asymmetries.
The essence of the alternative approach is that this low entropy
boundary condition is actually the only asymmetry we need. There
need be no additional time-asymmetric generalization at work in
nature (at least in its thermodynamic manifestations), in order to
account for the observed phenomena. This one-asymmetry
approach originates in the late 1870s, in Boltzmann's response to
Loschmidt's 'reversibility objections'. In essence, I think—
although he himself does not present it in these terms—what
Boltzmann offers is an alternative to his own famous H-Theorem.
The H-Theorem offers a dynamical argument that the entropy of a
non-equilibrium system must increase over time, as a result of col-
lisions between its constituent molecules. As Burbury made clear in
the 1890s (see Section 7), the time-asymmetry of the argument
stems from a time-asymmetric independence assumption. Roughly,
it is assumed that the velocities of colliding particles are indepen-
dent before they interact.
The statistical approach does away with this dynamical argument
altogether. In its place it offers us a simple statistical consideration.
In a natural measure on the space of possible microstates of a phys-
ical system, microstates which are such that the system approaches
equilibrium over time vastly outnumber those in which it behaves
in other way in the set of all microstates compatible with a give
macrostate). In a sense, this is not so much an explanation for the
behaviour of the system as a reason why no special explanation is
necessary—a reason for thinking the behaviour unexceptional.
As Boltzmann himself notes (1877, 193), the statistical consider-
ations involved in this argument are time-symmetric. For a system
in a given non-equilibrium macrostate, most microstates compati-
ble with that macrostate are such that the system equilibrates
towards the past, as well as towards the future. In practice, of
course, this is not our experience. The low entropy systems with
which we are familiar typically arise from systems of even lower
entropy in the past. For example, the temperature difference
between a cup of tea and its environment arises from the greater
temperature difference between the boiling kettle and its environ-

27
Huw Price

ment. Entropy seems to decrease towards the past, and Boltzmann's

time-symmetric statistics render this behaviour exceptionally puz-
zling—in making it unsurprising that the tea cools down, they make
it puzzling why there was ever such a concentration of heat in the
first place.
Taken seriously, Boltzmann's statistical approach thus directs our
attention to the fact that entropy was very low at some point in the
past. This 'boundary condition' is time-asymmetric, as far as we
know, but this is the only time-asymmetry in play, on this view. The
observed time-asymmetry of thermodynamic phenomena is thus
taken to arise from the imposition of an asymmetric boundary con-
dition3 on the time-symmetric probabilities of the Boltzmann mea-
sure. Diagrammatically:
Asymmetric boundary condition—entropy low in the past
+ Symmetric default condition—entropy likely to be high, ceteris paribus

Observed asymmetry.

3.3 Advantages of the one-asymmetry view

The immediate advantage of the one-asymmetry view is its relative
simplicity, or theoretical economy. If it works, it simply does more
(or as much) with less. The best response to such a claimed advan-
tage would be to show that rival views do something more—and
something worth doing, of course, for otherwise the extra expendi-
ture is for nought. In this area, unfortunately, one's conception of
what needs doing depends very much on very issue at stake.
Proponents of two-asymmetry views take it that there is a time-
asymmetric nomological generalization evident in thermodynamic
phenomena, and hence take explanation of this generalization to be
something worth paying for (in the coin of theoretical complexity).
But this consideration will not move their one-asymmetry rivals,
who deny that there is any such generalization at work.
Disputes of this kind are difficult to settle, because the two sides
have such different views of what counts as winning. The best
3
The use of the term 'boundary condition' here is intended to reflect
the role the proposition in question plays in the explanation of the ther-
modynamic asymmetry, relative to 'local' dynamical constraints on the
behaviour of matter. Thus, it reflects the view that the condition does not
flow from those constraints, but needs to be given independently. It is not
intended to exclude the possibility that the condition in question might
come to be treated as lawlike by some future physics.
28
 Burbury's Last Case: The Mystery of the Entropic Arrow

dialectical strategy is often to exploit one's opponents' own concep-

tion of what counts as losing—in other words, to show that their
view fails by their own lights. This is the strategy I shall follow
below, in arguing against the two-asymmetry view. First, however,
I want to respond to what may seem an objection to the one-
asymmetry approach.

3.4 A surprising consequence of the one-asymmetry view?

The probabilities involved in the statistical picture are time-sym-
metric. They imply that entropy is very likely to be high in the past,
as well as in the future. Yet entropy seems to have been extremely
low in the past. Doesn't this amount to a strong disconfirmation of
the view, by any reasonable standards? In other words, doesn't it
give us good grounds for thinking that the Boltzmann measure is
inapplicable to the real world?
It seems to me that there are two possible strategies available to
one-asymmetry view, in response to this objection:
(i) The no asymmetry strategy. This involves saying that the
Boltzmann measure is not erroneous, because the past low
entropy is just a fluke—the kind of possible though extremely
unlikely outcome explicitly permitted by the Boltzmann proba-
bilities. This sounds like simply thumbing one's nose at the
canons of confirmation theory, but Boltzmann himself suggests a
way of making this option less unappealing than initially it seems.
In an infinite universe, even very low entropy states may be
expected to occur occasionally, simply by random fluctuation. If
creatures like us can only exist in appropriate proximity to such
fluctuations, then it is not surprising that we find ourselves in an
otherwise unlikely kind of world.4 (Since this approach does away
altogether with the need for an additional time-asymmetric
restriction on the Boltzmann probability measure, it is genuinely
a 'no asymmetry' approach: on the global scale, it tells us, there is
no thermodynamic asymmetry!)
(ii) The ceteris paribus strategy. The second approach takes the
Boltzmann probabilities to be merely 'default' expectation values,
to be used in the absence of overriding factors—and the past low
entropy boundary condition, as yet not fully understood, seems
4
The closest Boltzmann seems to come to making this anthropic pro-
posal explicit is in his (1895, 415). I am grateful to Jos Uffink for pointing
out to me that it is not nearly as obvious that Boltzmann actually had this
point in mind as tradition has tended to assume.
29
Huw Price

to be one such factor. In other words, the Boltzmann measure

tells us what is likely to be the case, other things being equal—and
we know of one relevant respect in which other things are not
equal, namely, the low entropy past.
Note that the one-asymmetry approach cannot avail itself of a third
escape strategy, that of claiming that the Boltzmann probabilities
are reliable only in one direction. If the probabilities become asym-
metric in this way, we no longer have a one-asymmetry view.
Thus for the one-asymmetry approach the choice seems to be
between (i) and (ii). Option (i) is now widely regarded as unwork-
able. There are two major problems. The first is that if this sugges-
tion had been true, we should not have expected to find any more
order in the universe than was already known to exist, for that was
certainly enough to support us. The second is that it is much easier
for a fluctuation to 'fake' historical records, by producing them from
scratch, than by producing the real state of affairs (of even lower
entropy) of which they purport to be records. On this view, then,
historical records such as memories are almost certainly mislead-
ing.5
Option (ii) has a surprising consequence. Given that the
Boltzmann probabilities are time-symmetric, it implies that they
have must the same 'default' status towards the future as towards the
past, and hence that they cannot give us grounds for confidence that
entropy will not decrease in the future. All we can reasonably say is
that it is very unlikely to decrease, other things being equal—i.e., in
the absence of the kind of overriding factors which make it decrease
towards the past. In other words, statistical arguments alone do not
give us good grounds for confidence that the Second Law will con-
tinue to hold. The statistics cannot rule out (or even give us strong
grounds for doubting) the possibility that there might be a low
entropy boundary condition at some point in the future, as there
seems to be in the past. (In order to clarify this possibility, we need
to understand more about the reasons for past boundary condition.)

5
For more details, see Price 1996, ch. 2, and Albert 2000, ch. 4. I believe
that the point was made originally by C. F. von Weizsacker 1939, in a paper
which appears in English translation as §11.2 in von Weizsacker 1980. Von
Weizsacker notes that 'improbable states can count as documents [i.e.,
records of the past] only if we presuppose that still less probable states
preceded them.' He concludes that 'the most probable situation by far
would be that the present moment represents the entropy minimum, while
the past, which we infer from the available documents, is an illusion' (1980,
144-5).
30
 Burbury's Last Case: The Mystery of the Entropic Arrow

In other words, the one-asymmetry view leads us to the kind of

open-mindedness about the future thermodynamic behaviour of
the universe that Burbury himself seemed to recommend (for some-
what different reasons, as we'll see), in the passage with which we
began. In my view this is a major 'sleeping' consequence of
Boltzmann's statistical approach to the thermodynamic asymme-
try—'Boltzmann's Time Bomb', as I call it in Price 2002—which
has been triggered by the recent discovery of the cosmological ori-
gins of the low entropy past.
Some people may feel that this consequence of the one-asymme-
try view is a reason for preferring its rival—better time-asymmetric
probabilities than probabilities whose predictions can't always be
trusted. In my view, this reaction is a case of shooting the
messenger. If this is what our best theory of the origins of the ther-
modynamic asymmetry tells us, shouldn't we take it seriously,
rather casting around for an alternative theory? (It is not as though
we have any good independent reason to reject the conclusion.)
Why is the one-asymmetry approach the best theory? In part
because it is simpler, and in part because the alternatives are fatally
flawed. Justifying the latter claim brings me to the main task of the
paper, which is to lay out some objections to the two-asymmetry
approach.

4. Two Kinds of Two-asymmetry View

I want to raise two objections to the two-asymmetry view. Neither

objection applies to all versions of the two-asymmetry view, but all
versions are subject to one objection or other, and some versions are
subject to both. In order to make things clear, I need a simple tax-
onomy of possible views. For completeness, I include the one-
asymmetry view. I'll present the taxonomy as a decision tree, as in
Figure 1. Answers to a few simple questions lead to one to one or
other of the four possible positions I want to distinguish.
The first issue is whether we need a second asymmetry at all—
whether the explanation of thermodynamic phenomena requires a
time-asymmetric nomological generalization, in addition to low
entropy past boundary condition. One-asymmetry approaches deny
this, of course, and hence occupy position A on the tree.
Next, two-asymmetry approaches may usefully divided into two
broad (though non-exclusive) categories, depending on their view of
the source of the second asymmetry. One category comprises a
group of approaches which seek a source for the second asymmetry
31
Huw Price

The puzzle of the thermodynamlc time-asymmetry

One time-asymmetry Two time-asymmetries

A I

Initial randomness the key to Dynamical cause for the

the second asymmetry second asymmetry

Internal External External Internal

Internal inital randomness | Internal asymmetric dynamics

(e.g., the H-Theorem) Interventionism ( e -9- Albert's GRW proposal)
B C D
Figure 1. Four possible views of the origin of the themodynamic arrow.

in some asymmetric 'boundary condition', involving randomness,

independence, or 'lack of correlation', in the initial motions of the
microscopic constituents of matter. More precisely, these approaches
take the second asymmetry to rest on the fact that there is such ran-
domness or lack of correlations in the initial but not the corre-
sponding final conditions. Approaches of this kind may be further
sub-divided according to their answer to the following question. Is
the 'internal' randomness of the microscopic motions of the con-
stituents of a system sufficient (position B), or do environmental
influences play a crucial role, as held by interventionist approaches
(position C)?
The second broad category of two-asymmetry approaches com-
prises views that seek a dynamical cause for the second asymmetry—
in other words, some identifiable factor dynamical without which
entropy would not be constrained to increase. Again, views of this
kind may usefully be sub-divided according to their view of the
'location' of this cause. For some views it essentially involves an
external influence (position C). For others it is a feature of the inter-
nal dynamics of normal systems—i.e., it would normally be present
even in a completely isolated system (position D).
Interventionism thus turns up in both broad categories. On the
32
 Burbury's Last Case: The Mystery of the Entropic Arrow

one hand, the interventionist's external influences comprise an

identifiable cause, without which (according to the interventionist)
entropy would not necessarily increase. On the other, it turns out to
be crucial that these external influences are suitably random, uncor-
related with the internal motions of the systems on which they exert
an influence.
Interventionism thus provides the obvious overlap between the
two broad second-level categories of two-asymmetry approaches
identified in Figure 1. In principle, these categories might also turn
out to be non-exclusive for a more subtle reason, namely that 'ini-
tial randomness' itself might be held to be a dynamical cause for the
second asymmetry. This possibility will be clarified below. For the
time being, I want to discuss causal approaches and initial random-
ness approaches separately. As I have said, I want to raise two objec-
tions to two-asymmetry approaches. The first objection is to causal
approaches, and hence applies to options C and D. The second is to
the initial randomness approach, and hence applies to options B and
C. Interventionism at least thus turns out to be doubly at fault, by
my lights, being vulnerable to both objections. And if there is a fur-
ther region of overlap between the causal and initial randomness
approaches, the same will be true there.

5 The Counterfactual Confinement Problem

Causal approaches seek a dynamical factor responsible for the gen-
eral tendency of entropy to increase—some factor without which
entropy would not increase, at least with the observed regularity. As
we noted, this factor needs to be time-asymmetric, for otherwise
entropy would be constrained to be constant (non-decreasing in
both directions).
One version of such an approach is interventionism, which takes
the cause to be provided by influences from the external environ-
ment 'coming at haphazard' (in the words of Burbury (1894, 320),
who seems to have been one of the first to suggest this idea). For the
moment, let's ignore the role of 'haphazardness', or randomness,
and focus on the external nature of these influences. I want to call
attention to the implied counterfactual claim:
(5a) If there were no such external influences, the observed phe-
nomena would be different—entropy would not increase
monotonically, in the observed fashion.

33
Huw Price

This counterfactual seems part and parcel of what it means to say

that the observed increase in entropy is caused by such external
influences.
If we accept that we need a dynamical cause for observed monot-
onic increase in entropy, but reject interventionism, then the alter-
native is to locate the cause in some asymmetric feature of the inter-
nal dynamics of matter. A recent example of such a view is David
Albert's (1994, 1998, 2000) suggestion that the collapse mechanism
in the GRW interpretation of quantum theory provides such an
asymmetry. The details of this proposal need not concern us here,
but again I want to call attention to the implied counterfactual:
(5b) If there were no such asymmetric mechanism, the observed
phenomena would be different.
In his 1994 paper, Albert accepts this counterfactual claim, at least
implicitly, in taking seriously the objection that if his suggestion
were correct, entropy would not increase in systems containing too
few constituents to allow the GRW mechanism to have its effect:
[T]he collapse-driven statistical mechanics ... will entail that an
extraordinarily tiny and extraordinarily compressed and absolutely
isolated gas will have no lawlike tendency whatever to spread out.
It can hardly be denied, therefore, that runs strongly counter to
our intuitions, [sic]
What it does not run counter to, however (and this is what has
presumably got to be important, in the long run) is our empirical
experience. (1994, 677)

In a later paper, Albert accepts the counterfactual claim explicitly,

saying that it is
perfectly right ... that anybody who claims that one or another
causal mechanism called M is what actually underlies the tenden-
cies of the entropies of thermodynamic systems to increase must
also be claiming that if that mechanism were not operating then
would be no such tendencies. (1998, 16)
Let us think about what this means. In a deterministic world, on
this view, thermodynamic systems would not behave as they nor-
mally behave. Deterministic coffee would not grow cold,
Newtonian cider would not need bottling. More precisely, if it were
true that the observed thermodynamic behaviour is caused by some
asymmetric indeterministic mechanism in the dynamics, then a

34
 Burbury's Last Case: The Mystery of the Entropic Arrow

means for turning off this mechanism would be a means for turning
off the general tendency to equilibration. Useful as this might prove
in practice, our present concern is with plausibility. Do we have
good grounds for thinking that such counterfactuals are true?
Similarly in the interventionist case. Is it really plausible that tem-
peratures would not equalize in a completely isolated laboratory?
In the remarks quoted above, Albert says that these counterfac-
tuals do not 'run counter to ... our empirical experience', and on the
assumption that the mechanism in question (GRW, or random
external influences) does obtain in the world of our empirical expe-
rience, this is quite true. But it is true in the trivial sense that in
virtue of having an antecedent which is contrary to fact, every strictly
counterfactual conditional does not run counter to empirical experi-
ence. For strict counterfactuals especially, then, not running counter
to experience is a very long way from being supported by experience
(and, pace Albert, the latter is what has really got to be important,
in the long run).
Albert also suggests that the issue is whether a gas has a 'lawlike
tendency ... to spread out', and this seems to me to be a little mis-
leading. The lawlike character of what we observe is disputable.
Indeed, it is precisely the point at issue. As I have noted, one-asym-
metry approaches to the thermodynamic arrow differ from their
two-asymmetry rivals mainly in denying that there is any asymmet-
ric nomological generalization or disposition of matter to be
explained. The real issue is not whether the gas would have a law-
like tendency to spread out, but simply whether it would spread out.
It is not enough for these causal views to claim that a world without
the cause in question would differ solely in modal respects.
These approaches thus require that the causal factor 'makes a dif-
ference'—that it alters the behaviour of matter in some way. Note
especially the burden of proof. It is the proponents of these causal
views who need a justification for the counterfactual claims (5a) and
(5b). Their opponents need only say that they see no reason to
accept such counterfactuals. Hence it is not sufficient for the causal
view to object that its opponents have no basis for a contrary coun-
terfactual claim. Thus Albert 1998 responds to the present objec-
tion as follows: 'If the GRW-theory is right, then there simply are
no [Boltzmann-like probability distributions over initial condi-
tions]; and so (of course) there would be no such distributions to fall
back on in the event that one were to entertain a counterfactual sort
of GRW-theory with the spontaneous localizations removed.'
However, my point is that it is Albert who needs to justify the coun-
terfactual (5b), not his opponents who need to refute it.
35
Huw Price

This objection seems to me to be very powerful, and it has a long

history, at least as applied to interventionism. I'm not sure who first
noted that it seemed implausible to claim that temperatures would
not equalize, or gas not disperse, in a genuinely isolated system, but
the point is certainly well known. Yet interventionism continues to
attract supporters. This sort of situation is common in science and
philosophy, of course, but it is often a sign that the two contending
theories are operating with different conceptions of the nature of
the problem. One way forward is to try to clarify the common task—
to step back from the various proposed solutions, and ask what the
problem actually is.
The debate about the asymmetry of thermodynamic phenomena
seems to me very much in need of this kind of clarification. The
approaches I have criticized in the section—interventionism, and
the asymmetric internal dynamics view—seem to me to be take for
granted that thermodynamic phenomena present us with a distinc-
tive lawlike regularity, for which we need to find a basis. Once this
is taken for granted (and it is accepted that the Second Law of
Thermodynamics cannot simply be a primitive law, inexplicable in
terms of any more basic feature of matter), it implies that there is
something more basic, without which the regularity in question
would not obtain. The task then is just to discover what that 'some-
thing' actually is. Against this background, the present objection is
bound to seem more of an issue of detail than an objection of prin-
ciple.
However, the background assumption is not compulsory. There is
an alternative view of the phenomena, according to which equili-
bration towards the future is not a nomological regularity. On the
contrary, it is the product of a weaker, time-symmetric, default con-
dition on the behaviour of matter, in combination with a single
time-asymmetric constraint (the low entropy past boundary condi-
tion). On this view—the one-asymmetry view—what happens
towards the future is simply not the manifestation of a time-asym-
metric nomological constraint.
Of course, what happens in the other direction—towards the
past—is not in accordance with the time-symmetric default condi-
tion. Hence it does call for explanation, on this view. Thus there is
a time-asymmetry in the phenomena, something puzzling in the
behaviour of matter—but the puzzle is the way it behaves towards
the past, not the way it behaves towards the future. In other words,
the one-asymmetry approach takes interventionism and the asym-
metric internal dynamics view to be trying to answer a question
which doesn't need addressing.
36
 Burbury's Last Case: The Mystery of the Entropic Arrow

It would be naive to hope that this kind of second-order point

will be suddenly convincing, where the first-order objection has
failed for so long. However, it does seem to me important to clarify
the debate in this way. To a considerable extent, rival approaches to
the thermodynamic asymmetry are not offering different answers to
the same question, but answers to different questions. In order to
make progress, we need first to recognize these competing concep-
tions of the nature of the problem—competing views of what an
answer to the puzzle of the thermodynamic time-asymmetry would
look like. In particular, we need to distinguish between the one-
asymmetry and two-asymmetry conceptions of the nature of the
task.

6. The Use and Abuse of Initial Randomness

I now turn to those versions of the two-asymmetry approach which

attribute the general tendency of entropy to increase to a character-
istic of the initial microstates of interacting matter—to 'initial ran-
domness', or 'lack of initial correlations'. (In Price 1996 I called it
the 'Principle of the Independence of Incoming Interactions', or
PP.) Again, this assumption needs to be time-asymmetric, applying
to initial conditions but not to final conditions. Otherwise, it yields
no asymmetry.
Figure 1 gave us two kinds of example of such views. One was
interventionism, which relies on the assumption that the incoming
influences exerted on a thermodynamic system by the external envi-
ronment are uncorrelated with the microstates of the system itself.
The incoming influences are assumed to 'come at haphazard'. In
this case the time-asymmetry is held to reside in the fact that the
outgoing connections to the external environment do not 'go at
haphazard', for otherwise, as just noted, the argument yields no
asymmetry. In other words, it needs to be held that the system is
correlated with its environment after their interaction, in a way in
which it is not correlated beforehand.
The second version of this initial randomness approach is the one
exemplified by the grandfather of would-be derivations of the
Second Law, the H-Theorem itself. This differs from intervention-
ism in that the initial randomness invoked is essentially an 'internal'
matter, not dependent on a contribution from the environment.
Burbury's great contribution to the discussion of the H-Theorem in
the 1890s was to put his finger on the time-asymmetric assumption
in question. It is the famous 'assumption of molecular chaos', or the
37
Huw Price

stofizahlansatz—roughly, the principle that the velocities of inter-

acting molecules are independent before they collide. As we shall see
in a moment, it was Burbury who pointed out why the correspond-
ing condition could not be expected to hold after collisions.
My discussion of these initial randomness approaches is in two
parts. I begin (Section 7) by examining the historical and theoreti-
cal basis of some assumptions typically associated with this
approach, from which it derives much of its appeal. This examina-
tion reveals a surprising foundational lacuna, apparently obscured
from view, at least in some quarters, by some longstanding errors of
interpretation of early arguments in the field (including especially
those of Burbury himself).
In the second stage, in Section 8, I want to treat the initial
randomness approach in a more abstract way, paying particular
attention to issues of explanatory and epistemological structure. I
distinguish several possible forms the approach may take, and argue
that all forms fall victim to one or other of several sub-problems.
The main distinction is between (i) those versions of the approach
which claim that our grounds for believing the entropy will contin-
ue to increase are that we have prior justification for believing in
initial randomness (in some appropriate form); and (ii) those which
take the epistemology to run the other way—which take the relevant
initial randomness principle to be known by inference from an inde-
pendently justifiable belief that entropy will continue to increase. I
argue that in different ways, both versions suffer from the problem
that their assumed epistemology is ungrounded—the required pri-
mary beliefs are simply not justifiable, by ordinary scientific stan-
dards. The latter approach also suffers from another problem. The
initial randomness assumption it invokes turns out to come in weak-
er and stronger forms. In weaker forms the proposed explanation of
entropy increase turns out to be question-begging in its epistemol-
ogy, and to introduce no genuine second time-asymmetry, of the
kind the two-asymmetry view requires. In stronger forms it runs
out to be vulnerable to the counterfactual containment problem.
These objections may seem surprising, for the initial randomness
assumptions have an apparent naturalness and plausibility. This is
strikingly exemplified in what for more than a century has often
been seen as one of the key problems in the thermodynamics of
non-equilibrium systems—that of dealing with objections to the H-
Theorem which turn on the claim that the assumption of molecular
chaos cannot continue to hold over time, because the molecules of a
gas become correlated as they interact. On this view, as Ridderbos
puts it,
38
 Burbury's Last Case: The Mystery of the Entropic Arrow

the central problem in non-equilibrium statistical mechanics [is

that] equilibrium can only be obtained and entropy can only
increase for those systems for which an appropriate analogue of
the StoBzahlansatz can be shown to hold. That is, a necessary
condition for the approach to equilibrium is that the system has
to get rid of correlations which are continually being built up
dynamically by the interactions between the constituents of the
system. (1997,477)
A number of authors—e.g., Bergmann and Lebowitz (1953), Blatt
(1959), Ridderbos and Redhead (1998)—have suggested that inter-
ventionism provides a solution to this 'central problem'. Where
does the time asymmetry come from, on this view? From the fact, it
is claimed, that the incoming influences 'from the environment' are
uncorrelated, whereas the corresponding 'outgoing influences' are
correlated, in virtue of the interaction between system and environ-
ment. As Ridderbos and Redhead put it:
This is why the argument cannot be applied in the reverse time
direction to argue that equilibrium will be approached into the
past; in the ordinary time direction the 'incoming' influences are
the influences from the environment on the system, and these are
uncorrelated, but in the reversed time direction the 'incoming'
influences are the influences the system exerts on its environment
and these will be correlated. (1998, 1261)
This tradition thus seems committed to the following two princi-
ples:
(6a) Particle motions are independent before interactions—i.e., if the
particles concerned have not interacted in their common past.

(6b) Interactions give rise to (or in general increase) correlations.

These principles are logically independent, in the sense that neither
implies the other. Yet each is clearly time-asymmetric, and neces-
sarily so, if both are to be maintained. For each principle is incom-
patible with the time-inverse of the other.
On the face of it, moreover, neither of these principles depends
on the low entropy past. Thus, remarkably, this tradition seems to
be committed to the existence of three independent time-asymme-
tries: (6a), (6b), and the low entropy past boundary condition. And
the combination is of doubtful coherency. After all, if there is a
dynamical reason for the accumulation of correlations in one direc-
tion due to interaction, and (as typically assumed in these discus-

39
Huw Price

sions) the dynamics is time-symmetric, then how could it fail to be

the case that correlations accumulate in both directions? Puzzling
assumptions, then, but as we have seen, also very natural ones to
contemporary intuitions.
How did the tradition get to this point? Where did this peculiar
combination of assumptions come from? It turns out that in order
to answer these questions, we need to return to Samuel Burbury's
contribution to the debate about the H-Theorem in the 1890s.

7. Origins of an Orthodoxy

In his initial (1894, 1895) contributions the debate of the 1890s,

Burbury points out that Boltzmann's proof of the H-Theorem
relies on what Burbury calls 'Condition A' (later 'Assumption A).
This amounts to the assumption that the velocities of colliding par-
ticles are independent, before collisions. Burbury also points out
that if this condition applied also to reverse motions, the H-
Theorem would imply that the gas in question would also be
approaching equilibrium in reverse, and this could only be true if
the gas were already at equilibrium. So in a case in which the gas is
not in equilibrium, and in which Assumption A does hold initially,
the assumption is not true of the reversed motions at the end of the
interval in question.
After Burbury's initial presentation of the point in his (1894) and
(1895) letters to Nature, the argument is reproduced in many places
both by Burbury himself (e.g., Burbury 1899, §39) and by others. In
particular, it is reproduced by the Ehrenfests in their important sur-
vey article in 1912 (Ehrenfest and Ehrenfest 1959, 40; 85, n. 65).
Roughly stated, Burbury's observation might seem to amount to
the principle that particle motions are not uncorrelated after colli-
sions—i.e., to principle (6b). It seems to be interpreted this way by
Ridderbos, among others:
It was the Ehrenfests in their famous 1912 article [Ehrenfest and
Ehrenfest 1959] who in a careful analysis of the StoBzahlansatz
pointed out the asymmetry which it contains; in general a distri-
bution of molecules in a gas which satisfies the StoBzahlansatz at
an initial moment will fail to do so at a later moment as a result of
the correlations between the molecules which are built up dynamical-
ly as a result of the collisions. (1997, 526, emphasis added.)
This interpretation is simply mistaken, however. The Burbury,
Ehrenfest and Ehrenfest (BEE) argument that the stofizahlansatz
40
 Burbury's Last Case: The Mystery of the Entropic Arrow

(or Assumption A) does not hold after a period of interaction relies

on two assumptions: (i) that the gas in question is not in equilibri-
um, and (ii) that the condition does obtain at the beginning of the
time interval in question. Given these assumptions, the H-Theorem
implies that entropy increases over the time interval in the forward
time sense, and hence that it decreases in the reverse time sense.
From the latter fact, a second application of the H-Theorem, and
modus tollens, we may infer that Assumption A does not obtain at the
end of the time interval.
Note the crucial role of the assumption of initial non-
equilibrium. Far from revealing a general tendency for correlations
to accumulate as a result of collisions, the BEE argument turns on
nothing more than this: Assumption A fails in the reverse motion
because entropy was low in the past. Not only is this not a result of
interactions; it seems independent of whether there actually are any
interactions!
The BEE argument thus supports neither (6a) nor (6b). It pro-
vides no basis for the view that there is an asymmetrical tendency
for interaction to give rise to correlations in the 'forward' time
sense, as (6b) asserts. And it gives us no a priori justification for
assuming that Assumption A does hold at any given 'initial'
moment, as (6a) maintains. On the contrary, in fact. Since our rea-
son for thinking that Assumption A does not hold later is that we
know that entropy was low earlier, we might think—by symmetry,
as it were—that we are not justified in assuming with (6a) that
Assumption A holds earlier, until we know whether entropy is low
later. Clearly, this thought has the potential to undermine any pro-
posal to make (6a) our reason for thinking that entropy will not
decrease in the future. (More on this in a moment.)
It is worth noting that Burbury himself seems to regard
Assumption A as unrealistic, and indeed, in the remark I quoted at
the beginning, offers what seems a lone voice of caution about our
right to assume that entropy will continue to increase. However,
even the astute Burbury does not seem to doubt that we are justified
in assuming (6a) for the case of incoming influences from an exter-
nal environment. His caution about Assumption A seems to be
based on the intuition that interactions within all but an ideally
rarefied gas will induce violations of Assumption A (due to the pro-
duction of 'velocity streams', for example). He does not seem to see
that there is a more basic puzzle about the origin of the asymmetry
in the assumption of independence for initial but not for final
motions.
The BEE argument is not the only reason for thinking that the
41
Huw Price

stofizahlansatz cannot continue to hold over an extended time peri-

od. A more basic reason is first noted, so far as I am aware, by J. H.
Jeans:
The effect of this assumption [i.e., Assumption A] is to enable us
to regard certain probabilities at any given instant as indepen-
dent, and we then assume not only that the probabilities at a later
instant are inter se independent, but also that they are indepen-
dent of the events which took place at any earlier instant. This
assumption cannot be logically reconciled with the fact that the
motion of the system is continuous in time, i.e., that the events
which occur at any instant depend on those which occurred at a
previous instant. (1903, 598)
Jeans calls this an a priori argument that Assumption A cannot con-
tinue to hold over a time interval. He then goes on to give a second
argument, which he terms a posteriori. This latter argument is
essentially the BEE argument. And Burbury, in turn, immediately
adopts Jeans' a priori argument, endorsing it, for example, in (1903,
530) and (1904).
Does Jeans' a priori argument support (6a) and (6b)? Not at all, for
it is not time-asymmetric. Nor incidentally, does it depend on the
assumption that the constituents of the system in question actually
interact. (As I noted, this is also true of the BEE argument.) In other
words, Jeans' logical point cannot be the basis of a one-way tendency
for correlations to be built up dynamically as a result of collisions.
I know of one more argument which might at first sight seem to
support (6a) and (6b). Consider two particles, of initial momenta p
andp'. If they interact with each other (and with nothing else), con-
servation of momentum ensures that their combined momentum
after collision is p + p'. Doesn't this imply that the two particles are
now correlated—that their momenta cannot vary independently?
However, a moment's reflection reveals that this simple argument
cannot provide a basis for (6b) and (6a). For one thing, there is
nothing time-asymmetric here. The argument applies equally well
in reverse. Indeed, it applies not only at any earlier or later time,
whether or not the particles collide; but also, in a sense, at the ini-
tial time itself. If we are given the total momentum, and told that
there are only two particles, then, relative to this information the two
momenta are not independent. The diachronic content of this argu-
ment is essentially that of Jeans' argument, viz., that in virtue of the
determinism of the dynamics, a choice of the dynamical variables at
any one time fixes those values at all other times (so long as the
system remains isolated, of course).
42
 Burbury's Last Case: The Mystery of the Entropic Arrow

Thus it seems to me that in so far as the tradition maintains that

there is some time-asymmetric sense in which correlations accumu-
late as a result of collisions, it is simply mistaken. There is no sense
in which the dynamics produces correlations in a forward time sense
in which it does not also do so in the reverse time sense. As a result,
the idea that the asymmetry of (6b) could bolster that of (6a)—
could show why (6a) doesn't hold in reverse—is also without foun-
dation. On the contrary, in fact. Our only relevant reason for think-
ing that there are correlations in final conditions are that we know
entropy was low in the past. Far from showing us that there is a
past-future asymmetry, this suggests that we cannot know whether
there is such an asymmetry—whether, as (6a) claims, there are no
corresponding correlations in initial conditions—until we know
whether entropy is low in the future.
Perhaps even more damagingly, this brief excursion into the his-
tory of the initial randomness approach suggests that it provides no
legitimate basis whatsoever for the nomological second asymmetry
required by the two-asymmetry approach. At best, the asymmetry
of micro-correlations is simply the first asymmetry re-described—
the asymmetry of macroscopic boundary conditions, characterized
in another way.
In order properly to evaluate these objections, however, it is nec-
essary to bring their target more sharply into focus. It turns out
that there are several subtle variants of the initial randomness
approach, differently sensitive to these points. In the next section I
want to look at the approach in abstract terms, paying close atten-
tion to matters of explanatory and epistemological structure—that
is, in particular, to the issue of what is being taken as providing
grounds for believing what. It turns out, I think, that while there
are possible versions of the initial randomness approach which are
not guilty of question begging, and which do genuinely introduce
a second time-asymmetry, these versions have other problems. For
one thing, they are subject to the counterfactual containment prob-
lem.

8 The Formal Structure of the Appeal to Initial Randomness

Let's begin with some terminology, to be read in conjunction with

Figure 2. The propositions HiEntp and HiEntf say that there is a
macroscopic condition of high entropy in the past and the future,
respectively. Ranf and Ranp represent the corresponding micro-
scopic conditions at the other end of time. Figure 2 depicts the
43
Huw Price

combinations of these propositions and their negations we ordinar-

ily assume to be true of the actual universe: ~>HiEntp and -"Raiif,
and HiEnt f and Ran p .
Thus the two shorter double-headed arrows in Figure 2 represent
logical implications (modulo the laws). Given the dynamical laws, a
high entropy past (HiEntp) would imply and be implied by the cor-
responding kind of randomness in the future (Ranf); and a high
entropy future (HiEntf) implies and is implied by the correspond-
ing kind of randomness in the past (Ran p ). (In particular, therefore,
Ran p is logically the weakest assumption that implies HiEntf.)

Ranf
'Past' < • 'Future1
Ran p -* • HiEntf

Figure 2. Macroscopic entropy and microscopic correlation

In the most abstract terms, then, the initial randomness approach

is committed to the idea that a high entropy future is a consequence
of, and is to be explained by, the fact that Ranp—that is, by the ran-
domness of the earlier microstates. Perhaps the true explanans is
some stronger principle of which Ran p is a consequence—more on
this option below—but at any rate the later high-entropy macrostate
is thought of as a consequence of the earlier microstate, and Ran p
is by definition the weakest condition which will do the trick.

8.1 Explanation and epistemology

The initial randomness approach thus takes Ran p to explain
HiEnt f . It also seems natural to take -"HiEntp to explain -"Ranf.
Note that the 'intuitive' direction of explanation is past-to-future,
in both cases. But this alignment in our explanatory intuitions
should not be allowed to mask an important difference in the
epistemological structures of the two claims, and hence a reason to
be suspicious of the former. The grounds for suspicion turn on the
issue of the justification the initial randomness approach takes us to
have to accept that HiEntf.
In the case of the claim that ->HiEntp explains -"Ranf the episte-
mology is relatively unproblematic. Our grounds for thinking that
->Ranf—i.e. that there are microscopic correlations in the future—
44
 Burbury's Last Case: The Mystery of the Entropic Arrow

are simply that we have observed that the universe is highly ordered
now and in the past. Epistemological inference thus follows the
explanatory arrow, and all is well because the past state of affairs—
the assumed explanans—is an observable matter.6
In the other case, however, the epistemology turns out to be
problematic. To see why, let us note first that there are two broad
possibilities. The first is that the epistemology does follow the
explanatory arrow (taken to be past-to-future), so that our grounds
for thinking that HiEntf are that we have independent reason for
thinking that the required microscopic randomness (Ranp) obtains
now (or in the past). The second is the converse possibility—our
grounds for thinking that Ranp obtains now and in the past are that
we have independent reason to believe that entropy does not
decrease in the future (from which past randomness follows as a log-
ical or at least abductive consequence). Let us call these possibilities
past-to-future inference and future-to-past inference, respectively. I
want to show that both are problematic, though for different
reasons.

8.2 Past-to-future inference?

This version of the initial randomness approach requires that we
have grounds for believing that Ranp obtains, other than by infer-
ence from the fact that it is required for HiEntf. In what could these
grounds consist?
The first option is that we might observe Ranp directly. However,
it is not difficult to see that this is not possible, at least in general.
Just as it becomes impossible, in practice, to detect the correlations
associated a system's low-entropy past—to tell, for example,
whether a particular equilibrium sample of gas had a specified non-
equilibrium state at a given earlier time—so it is impossible to make
such determinations with respect to the future. (Among the factors
which make this kind of observation impossible, in real systems, is
the fact that measurements would be required across an entire
spacelike hypersurface of the past light cone of the relevant segment
of future spacetime.) Thus there is no prospect whatsoever of
6
For present purposes I ignore the sceptical issue concerning our
entitlement to believe our apparent evidence for a low entropy past, stem-
ming from von Weizsacker's observation (see Section 3.4 above) that if
Boltzmann's probabilities are our guide, then it is much easier to produce
fake records and memories, than to produce the real events of which they
purport to be records. Albert 2000, ch. 4, provides an excellent recent
account of this sceptical point.
45
Huw Price

detecting directly the 'hidden' correlations which would be required

for entropy to decrease in the future—nor, for the same reasons, of
excluding them on observational grounds.
If our acceptance of Ranp does not rest on direct observational
grounds, then on what? Could there be some direct but non-obser-
vational reason for accepting Ranp? There seem to be two kinds of
argument on offer at this point. One tries to appeal to the intuition
above, that correlations are typically produced by interactions, and
'hence' that none are to be expected before interactions—incoming
influences can be expected to be independent. However, we have
seen that the history of the subject offers us no sound basis whatso-
ever for this appeal, or for the asymmetric principle about interac-
tions on which it relies.
The second kind of argument appeals to statistical considera-
tions. It claims that failure of Ran_ would require highly improba-
ble coordination among the microscopic motions of matter, in order
to give rise to ~"HiEntf. However, in the absence of an independent
reason for thinking that the statistics are time-asymmetric, we have
no reason to think that the correlations required to give rise to
~"HiEntf are any more unlikely than those required to give rise to
-"HiEntp. Since the latter obtains despite these statistical considera-
tions, we have very good reason to doubt the general reliability of
statistical reasoning in this context. (After all, it fails in one case out
of a possible two.)
It might be objected that we do have good reason—very good rea-
son—to think that the relevant statistical arguments are reliable
past-to-future, namely, that they perform so well in predicting the
past-to-future behaviour of many real systems. However, in so far
as it claims to offer us a reason to believe Ranp, this argument
involves an inductive step. In effect, it relies on the claim that we are
justified in expecting this statistical success to continue. As such,
the claimed inference to Ranp thus becomes indirect, and future-to-
past (on which more in a moment).
Thus it appears that the epistemology of the claimed explana-
tion of HiEntf by Ranp cannot follow the explanatory arrow.
Unlike in the case of the explanation of ~'Ranf by """HiEntp, our
grounds for accepting the future explanandum cannot be that we
have independent grounds to accept the past explanans. If we have
good reason to believe the supposed explanandum at all, then, it
must be on some basis more direct. Let us now turn to that possi-
bility.

46
 Burbury's Last Case: The Mystery of the Entropic Arrow

8.3 Future-to-past inference—a direct case for HiEntr?

In this version, the initial randomness approach is the view that we
are led to some sort of initial randomness principle by abductive
inference from the prior discovery that entropy will continue to
increase. It is the view that in some form, initial randomness is the
best explanation for this independently established fact about the
physical world.
I want to raise two kinds of objection to this proposal. The first
echoes Burbury's point, from the quotation with which we began.
We simply don't have particularly strong reasons for believing that
entropy will continue to increase, at least in the distant future.
However, although the sentiment is Burbury's, the present justifi-
cation for it depends on considerations of which Burbury could not
have been aware. It depends on our current understanding of the
nature of the past low entropy boundary condition, and hence on
modern cosmology.
Within the last thirty years or so, cosmology has offered us a plausi-
ble hypothesis as to the origins of the observed low entropy. Briefly,
everything seems to turn on the fact that matter was very smoothly dis-
tributed, soon after the Big Bang. This is a very highly ordered
condition for a system dominated by gravity, and seems to supply a
low-entropy 'reservoir' which supplies the entire thermodynamic gra-
dient we observe in our region. (This story is very well told by Penrose
1989. On its links to the present issue, see Price 1996, ch. 2, Price 2002.)
Given this reason to think that the low entropy past has cosmo-
logical origins, we also have some basis for thinking that the issue of
future low entropy also turns on cosmological issues. In particular,
if we are entitled to believe that entropy will not decrease in the
future, then it must be because we are entitled to believe—on cos-
mological grounds—that there is no future low entropy condition,
of the same kind as the past condition. There seem to be two possi-
ble routes to such a belief. One would be theoretical, and turn on the
question as to whether our best cosmological models excluded the
possibility (or at least the likelihood) of such a future condition.
The other would be observational, at least in part, and turn on the
issue of what present manifestations there might be of such a future
boundary condition. At present, both routes are best very inconclu-
sive. (I shall say more about the lattSr possibility in a moment.) As
it stands, then, Burbury's cautious agnosticism remains the appro-
priate attitude. In other words, it is very far from clear that there is
an explanandum (HiEntf) in need of the kind of explanans (Ranp)
this approach takes initial randomness to offer.

47
Huw Price

I shall also argue that even if we did have good reason for think-
ing that entropy will continue to increase, this wouldn't count in
favour of the initial randomness approach, which would still fall
victim to the following dilemma. Depending on what is meant by
initial randomness, it is either (i) a condition too weak to support
a non-trivial two-asymmetry approach to the thermodynamic
asymmetry; or (ii) a condition that, in being stronger than neces-
sary, is subject to an analogue of the counterfactual containment
problem.

8.3.1 Why HiEntf isn't presently observable

If entropy were to decrease on a large scale at some time in the
future, would that fact be detectable now? If we assume determin-
ism, then in one sense, of course, it would be observable in princi-
ple to a Laplacian ideal observer. As we have already noted, however,
there is no prospect whatsoever of this sort of microstate-based
observation being possible for real observers in real systems. If it
were possible, moreover, then it would count as direct observation
of Ran p , and hence as a basis for a past-to-future inference. The
possibility we have now to consider is different. It is that there
might be macroscopic evidence now as to whether entropy contin-
ues to increase in the future.
This possibility is not as bizarre as it might at first sound. There
seem to be quite good grounds for thinking that certain kinds of
future low entropy boundary conditions would have observable
consequences now—and hence that at least to some extent, we can
have observational evidence that those boundary conditions do not
obtain in the future. The kind of evidence concerned is very much
the kind of evidence we have of the low entropy past. Essentially,
the latter evidence consists in low entropy 'remnants'—in systems
that had not yet had time to reach equilibrium, following the low
entropy condition in the past. (Stars and galaxies are striking exam-
ples of such remnants.)7
In an exactly analogous way—and proceeding on the same
assumptions, applied in reverse—future low entropy conditions
might well be expected to have manifestations now. These would be
systems that, in the reverse time sense, have likewise not had suffi-
cient time to equilibrate.
This possibility has been discussed in the literature in connection
7
Again, I am ignoring von Weizsacker's sceptical difficulties
about inference to the past.
48
 Burbury's Last Case: The Mystery of the Entropic Arrow

with a proposal made by Thomas Gold 1962, that the universe

might be globally symmetric in entropy terms, with a low entropy
future endpoint. The question of the advanced effects of such a
boundary condition has been addressed in variations of the
Ehrenfest two-urn models, with two-time boundary conditions.
(See, for example, Cocke 1967 and the references in Schulman 1997,
156.) The upshot of these investigations is that such a boundary
condition at a time Tj is not expected to be detectable before
Tj — treiax, where treiax is the relaxation time of the relevant physical
processes. Beyond that relaxation time, the system behaves in a
manner macroscopically indistinguishable from one which lacks the
future boundary condition. (The predictions assume a statistical
measure that is symmetric, before application of boundary condi-
tions. However, the conclusion concerning detectability outside the
relaxation time seems likely to apply even in asymmetric back-
ground measures.)
Thus it seems that even a very large low entropy future boundary
condition, comparable in magnitude to that in our past, might be
undetectable now, so long as it is sufficiently far away. How far away
is sufficiently far away? The issue depends on the relaxation time of
the relevant real physical processes. This question has been dis-
cussed a little in recent years (e.g., in Gell-Mann and Hartle 1994;
see also Price 1996, ch. 4), but the relevant issues remain open. At
any rate, it is clear that at present there is no strong observational
case for saying that there isn't a low entropy boundary condition in
the future.
It might be thought that there is easier observational route to the
conclusion that Hi£ntf. Can't we simply get there by induction
from the observed present behaviour of matter? We observe that the
Second Law holds now, with great consistency. Isn't that a reason to
believe that it will continue to hold in the future? In other words,
doesn't induction give us our explanandum (and abduction then our
explanans)?
The problem is that induction depends on the assumption that
the future will be like the past in the relevant respect. It is therefore
question-begging to try to use it to exclude the contrary hypothesis.
In other words, induction is powerless to exclude a theoretically
well-motivated hypothesis to the effect that the future will not be
like the past. (This is simply a more localized version of Hume's
point, viz., that we can't use inductive methods to justify induc-
tion.) Indeed, as I noted in Section 3.1, the issue between the one-
asymmetry and two-asymmetry views seems to come down to that
as to whether there is a projectible (or induction-supporting) gener-
49
Huw Price

alization to the effect that entropy will continue to increase. There

is no prospect of settling this issue by induction.
8.3.2 Why initial randomness would be in trouble, even if HiEnt f
were observable
Suppose for a moment that we did have good theoretical or obser-
vational grounds for believing HiEntf. Would we be justified in
making an abductive inference to Ranp? Yes, at least on some con-
ceptions of explanation. Since Ran p is by definition equivalent to
HiEntf, given the relevant laws, it follows that Ran p is an initial
condition sufficient to imply HiEntf, given the laws. If this is held
to be sufficient for explanation, then Ran p can indeed explain
HiEntf.
Unfortunately for the two-asymmetry approach, however, such
an explanation does not conform to the two-asymmetry model.
Ran p implies HiEntf, but the implication is logical, not nomologi-
cal. So long as HiEntf is a factlike matter, so too is Ran p . After all,
the one-asymmetry view also accepts that Ran p . On this view, the
default expectation is for both Ran p and Ran f (or equivalently, both
HiEntf and HiEnt p ); and the single asymmetry is the constraint
which supplies """HiEntp (or equivalently ""Ranf).
Thus if abductive inference is to get us anywhere useful from
HiEntf, from the two-asymmetry approach's point of view, it has to
get us to something stronger than Ran p . Call such a stronger propo-
sition *Ran p . In effect, the initial randomness approach will then be
committed to the counterfactual
(8a) If not *Ran p , then (probably) not HiEntf.
For if HiEntf were held to be likely to occur anyway, even in the
absence of *Ran p , the supposition that *Ran p could hardly be
required to explain HiEntf.
But why should we accept (8a)? We know it is logically possible
that Ran p (and hence HiEntf) should obtain without *Ran p . (By
definition, Ran p is sufficient for HiEntf, and *Ran p is logically
stronger than Ran p , so failure of *Ran p does not entail failure of
Ran p .) This version of the initial randomness approach will there-
fore need to argue on non-logical grounds that if *Ran p failed, so
too would Ranp—the result would not simply be a world in which
Ran p obtained anyway.
The dialectical position of the initial randomness approach here
is the same as that of causal versions of the two-asymmetry theory.
As we saw in Section 5, those approaches are committed to the claim
that if the causal mechanism in question did not operate—if a
50
 Burbury's Last Case: The Mystery of the Entropic Arrow

system were truly isolated from its environment, if there were no

GRW collapse, or whatever—thermodynamic systems would not
behave as they are observed to behave. In effect, this amounts to the
idea that entropy needs to be prevented from decreasing—that the
initial condition of the universe is such that entropy will decrease,
unless something intervenes. The one-asymmetry approach chal-
lenges this idea, arguing that the observed phenomena can be con-
strued as the product of a single time-asymmetric constraint (the
low entropy past) on an otherwise symmetric space of possibilities.
Unless we are given some reason to rule out this conception of the
origin of the phenomena in question, the causal approach is not
entitled to its counterfactuals. While the one-asymmetry model
remains a viable possibility, in other words, we cannot be justified in
claiming that without the second asymmetry, the relevant phenom-
ena would have been different.
All this transfers to the present case. Given the supposition that
HiEntf, the one-asymmetry approach will propose that the result-
ing past-future asymmetry—the fact that HiEntf but not HiEnt_—
reflects only one-time asymmetry, that of the low entropy past
boundary condition. In order to make a case for (8a), and hence for
a second asymmetry in the form of *Ranp, the initial randomness
approach needs a reason to disallow this one-asymmetry proposal.
No such reason seems to have been forthcoming.

8.3.3 Summary: The case against nomological initial randomness

The argument against an appeal to a 'strong' form of initial
randomness may be summarized in this way. If *Ranp were to be
justified on theoretical grounds, these grounds would need to be
time-asymmetric (since otherwise they would equally give us
*Ranf, and hence HiEntp). A time-asymmetric theoretical assump-
tion of this kind might be justified if we already knew HiEntf, and
therefore sought an explanation of this known fact. But (i) this is
not our situation—on the contrary, we are still looking for some rea-
son to believe HiEntf. And (ii) even if we did have such reason,
*Ranp wouldn't be necessary, since we have no reason to think that
situation is not merely the result of symmetric Boltzmann proba-
bilities, subject to a single time-asymmetric boundary condition.

8.3.4 Is the argument too strong?

It might be objected that the last objection proves too much. If it
worked, wouldn't it show that nomological explanations are always
51
Huw Price

too strong? For isn't the possibility always open that the same phe-
nomena would have occurred without the law in question?
This is an interesting point, which raises issues which go far beyond
the scope of this paper. In my view, the counterfactual containment
argument does pose a general problem for some strongly realist con-
ceptions of the nomological realm. For present purposes, however,
what matters is something that seems to distinguish the use of argu-
ment in the thermodynamic context from possible general uses.
In the present context, we have a relatively clear alternative to the
two-asymmetry approach's conception of the origins of the relevant
thermodynamic phenomena. That is, we have clearly in view a
means by which the same phenomena could have arisen, without a
nomological asymmetry. Assuming a deterministic dynamics, all it
takes is for the initial microstate of the universe to be 'normal', in
the sense of Boltzmann's measure, in the space of possibilities com-
patible with the initial macrostate. Hence the onus is on the propo-
nent of a two-asymmetry view to convince us that without the
nomological asymmetry supposedly in question—GRW collapse,
external influence, *Ranp, or whatever—the initial microstate
would not have been of that kind. (As I noted in Section 5, there is
no corresponding onus on opponents of the two-asymmetry
approach to defend the Boltzmann measure. Proponents of the two-
asymmetry approach are obliged to justify the relevant counterfac-
tual, and hence to exclude the Boltzmann measure, but opponents
can simply afford to be agnostic.)
A familiar argument for realism about nomological necessity is
that without it, observed regularities become incredible coinci-
dences. The implicit counterfactual claim is that without real neces-
sities, the regularities in question would probably fail. Whatever we
think about the merits of this argument, it is better than the analo-
gous defence of nomological asymmetry in the thermodynamic
case. There may perhaps be a prima facie case that regularities in
general would fail without laws. Post-Boltzmann, there isn't even a
prima facie case that thermodynamic systems would equilibrate
differently without a nomological constraint. Boltzmann showed us
that there is at least one plausible measure on the space of possible
initial microstates under which the observed phenomena are highly
likely, even in the absence of a nomological constraint. In doing so,
he placed the ball squarely in the court of defenders of nomological
asymmetry. To make their view stick—to return the ball—
defenders of that view need a reason to rule out Boltzmann's
measure. So far as I know, there has been no serious, let alone
successful, attempt to do so. (Once again, the point is not that
52
 Burbury's Last Case: The Mystery of the Entropic Arrow

Boltzmann's measure is a priori right, but simply that it is a serious

contender, which has not been shown to be wrong.)

8.4 Conclusion
When disambiguated, the initial randomness approach turns out to
be unable to provide a satisfactory explanation of the second asym-
metry, of the kind the two-asymmetry approach requires. The
faults of the approach depend on the precise form in which it is pre-
sented to us.
If it is offered in the spirit of a justification for believing that
entropy will continue to increase, then it fails because there is no
good reason to believe that the required condition of initial ran-
domness does obtain. (It is not observable, and statistical arguments
are unreliable, by parity of reasoning.)
If it is offered to us in the spirit of an inferred explanation of
something independently determinable, then again the approach
fails. In this case, the problem is both that the explanandum is not
independently justifiable, and that even if it were, the proposed
explanans would be either (i) too weak for the two-asymmetry
approach, in not supplying a nomological second asymmetry, or (ii)
unnecessary.

9. Starting in the Wrong Place

The failings of the two-asymmetry approach seem to boil down to

two points. The first is an epistemological failing, a vulnerability to
a modern version of Samuel Burbury's objection. It simply hasn't
been adequately established that on a global scale, the relevant phe-
nomena are as the two-asymmetry approach takes them to be. That
is, it hasn't been established that entropy will continue to increase.
We saw that what has been added since Burbury's time is the cos-
mological case for regarding this as an open question. Burbury was
rightly suspicious of the crucial assumption (the assumption that he
himself had first identified) of the leading argument for thinking
that entropy will continue to increase. The cosmological connection
supplies an independent reason for holding open the possibility that
it will not do so.
The second failing is methodological. Even if we grant the rele-
vant phenomena, the two-asymmetry approach has failed to estab-
lish that the explanation of those phenomena requires something
either nomological or time-asymmetrical (in addition, that is, to the
low entropy past boundary condition). To make this case, the two-
53
Huw Price

asymmetry approach needs to defend the kind of counterfactual we

identified at (5a), (5b) and (8a), and thus to exclude the alternative
one-asymmetry conception of the origin of the asymmetric phe-
nomena in question. No such defence has been offered.
In my view, these two failings are symptoms of a more basic fail-
ing, that of asking the wrong question to start with. The two-asym-
metry approach begins with the question: Why does entropy go up
towards the future—why is matter engaged in this 'uphill' journey?
The one-asymmetry approach begins instead with the question:
Why does entropy go down towards the past—why does the matter
begin its journey at such a 'low' spot? It is true that by one-asym-
metry lights these are in a sense the same question: the journey is
uphill because it starts in a low place. All the same, the latter version
puts the emphasis in the right place—it brings to the foreground the
crucial puzzle.
By two-asymmetry lights, the questions are not the same. A rea-
son why matter cannot go 'downhill' is not automatically a reason
why its journey must start at a low place. Hence the need for two
asymmetries, in this picture—and so much the worse for this
approach, as we have seen.
In my view, a great but still under-appreciated consequence of
Boltzmann's contribution to our understanding of the thermody-
namic asymmetry is that it directs our attention 'backwards' in this
way. It show us that if our interest is in the time-asymmetry, the
proper focus of our attention is the condition of the universe in
what we regard as its past, rather than some intrinsically time-
asymmetric characteristic of its current journey.
Thus the crucial question is why entropy is so low in the past, and
as noted in Section 8.3, modern cosmology has given this issue a
remarkably concrete form. But this question arises from an even
more basic one, which isn't essentially time-asymmetric: Why isn't
entropy almost always high, as the time-symmetric Boltzmann mea-
sure would lead us to expect? We'll still need to answer this latter
question, even if—as we currently have no very strong reason to dis-
believe, in my view—entropy turns out to decrease in the distant
future, and the 'end' of the universe is as peculiar as its 'beginning'.8

8
Early versions of some of this material were presented at a conference
in Groningen in September, 1999, at the Royal Institute of Philosophy
Conference at LSE in September, 2000, and in talks in Utrecht, College
Park and Tucson in April, 2001. I am grateful to participants for discus-
sions on those occasions; and indebted to David Atkinson and Craig
Callender for comments on earlier drafts, amongst much else.
54
 Burbury's Last Case: The Mystery of the Entropic Arrow

References
Albert, D. 1994. 'The Foundations of Quantum Mechanics and the
Approach to Thermodynamic Equilibrium', British Journal for the
Philosophy of Science, 45, 669-677.
1998. 'The Direction of Time', paper presented to the Annual
Meeting of the Eastern Division of the APA, Washington, D.C.,
December, 1998.
Bergmann, P. G. and Lebowitz, J. L. 1955. 'New Approach to Non-equi-
librium Processes', Physical Review, 99, 578-87.
Blatt, J. M. 1959. 'An Alternative Approach to the Ergodic Problem',
Progress in Theoretical Physics, 22, 745-56.
Boltzmann, L. 1877. 'Uber die Beziehung zwischen des zweiten
Hauptsatze der mechanischen der Warmetheorie' ('On the Relation of a
General Mechanical Theorem to the Second Law of
Thermodynamics'), Sitzungsberichte, K. Akademie der Wissenschaften in
Wien, Math.-Naturwiss, 75, 67-73 (reprinted in Brush 1966).
1895. 'On Certain Questions of the Theory of Gases,' Nature, 51,
413-15.
1964. Lectures on Gas Theory, Berkeley: University of California Press.
Brush, S. 1966. Kinetic Theory. Volume 2: Irreversible Processes, Oxford:
Pergamon Press.
Burbury, S. H., 1894. 'Boltzmann's Minimum Function', Nature, 51, 78.
1895. 'Boltzmann's Minimum Function', Nature, 51, 320.
1899. The Kinetic Theory of Gases, Cambridge: Cambridge
University Press.
1903. 'Mr J. H. Jeans' Theory of Gases', Philosophical Magazine,
Series 6, 6, 529-35.
1904: 'On the Theory of Diminishing Entropy', Philosophical
Magazine, Series 6, 8, 4349.
Cocke, W. 1967. 'Statistical Time Symmetry and Two-Time boundary
Conditions in Physics and Cosmology', Physical Review, 160,
1165-1170.
Culverwell, E. 1890a. 'Note on Boltzmann's Kinetic Theory of Gases, and
on Sir W. Thomson's Address to Section A, British Association, 1884',
Philosophical Magazine, 30, 95-99.
1890b. 'Possibility of Irreversible Molecular Motions', Report of the
British Association for the Advancement of Science, 60, 744.
1894. 'Dr. Watson's Proof of Boltzmann's Theorem on Permanence
of Distributions', Nature, 50, 617.
Ehrenfest, P. and Ehrenfest, T. 1959. The Conceptual Foundations of the
Statistical Approach in Mechanics. English translation by M.J.
Moravcsik (Ithaca, NY: Cornell University Press). Dover edn. 1990
(NewYork: Dover Publications).
Gell-Mann, M. and Hartle, J. 1994. 'Time Symmetry and Asymmetry in
Quantum Mechanics and Quantum Cosmology,' in Halliwell, Perez-
Mercader, and Zurek (1994), pp. 311-45.
55
Huw Price

Gold, T. 1962. 'The Arrow of Time,' American Journal of Physics, 30

403-10.
Gold, T. (ed.) 1963. The Nature of Time. Ithaca: Cornell University Press.
Hall, E. H. 1899. Review of S. H. Burbury, The Kinetic Theory of Gases
(Cambridge: Cambridge University Press, 1899), Science, New Series,
10, 685-88.
Halliwell, J., Perez-Mercader, J. and Zurek, W. (eds), 1994. Physical
Origins of Time Asymmetry, Cambridge: Cambridge University Press.
Jeans, J. H. 1903. 'The Kinetic Theory of Gases developed from a New
Standpoint', Philosophical Magazine, Series 6, 5, 587-620.
Lebowitz, J. 1993: 'Boltzmann's Entropy and Time's Arrow', Physics
Today, 9:93, 32-8.
Price, H. 1996. Time's Arrow and Archimedes' Point: New Directions for the
Physics of Time, New York: Oxford University Press.
2002: 'Boltzmann's Time Bomb', British Journal for the Philosophy of
Science, 53, 83-119.
Ridderbos, T. M. 1997. 'The Wheeler-Feynman Absorber Theory: A
Reinterpretation?', Foundations of Physics Letters, 10, 473-86.
Ridderbos, T. M. and Redhead, M. 1998. 'The Spin-echo Experiments
and the Second Law of Thermodynamics', Foundations of Physics, 28,
1237-70.
Schulman, L. 1997. Time's Arrows and Quantum Measurement,
Cambridge: Cambridge University Press.
Sklar, L. 1995. 'The Elusive Object of Desire: in Pursuit of the Kinetic
Equations and the Second Law', in Savitt, S., (ed.), Time's Arrows
Today., Cambridge: Cambridge University Press, 191-216. Originally
published in Fine, A. and Machamer, P., (eds), PSA 1986: Proceedings
of the 1986 Biennial Meeting of the Philosophy of Science Association,
vol. 2.
von Weizsacker, C. 1939. 'Der zweite Hauptsatz und der Unterschied von
der Vergangenheit und Zukunft,' Annalen der Physik (5 Folge), 36,
275-83.
1980. The Unity of Nature, New York: Farrar Straus Giroux.

56
Zeno's Arrow and the Significance of
the Present
ROBIN LEPOIDEVIN

I. Introduction
Perhaps the real paradox of Zeno's Arrow is that, although entirely
stationary, it has, against all odds, successfully traversed over two
millennia of human thought to trouble successive generations of
philosophers. The prospects were not good: few original Zenonian
fragments survive, and our access to the paradoxes has been for the
most part through unsympathetic commentaries. Moreover, like its
sister paradoxes of motion, the Arrow has repeatedly been dis-
missed as specious and easily dissolved. Even those commentators
who have taken it seriously have propounded solutions with which
they profess themselves to be perfectly satisfied. So my question is:
will Zeno's Arrow survive into the millennium just begun?
I certainly hope so. What I want to do in this paper is argue, not
simply for its preservation, but also for its creative reconstruction.
Every generation needs to reinvent the wheel. Arguments, conun-
drums, paradoxes have to be rediscovered and re-presented in the
contemporary idiom. Philosophy of time has of course moved on
since Zeno's day, but that is not to say that the Arrow cannot speak
to the kinds of issues that are now engaging us.
At the end of a characteristically detailed and illuminating com-
mentary, Jonathan Barnes offers the following observation:
It is usually supposed that Zeno's paradox carries with it some
philosophical theory about the nature of time; and Zeno's com-
mentators regularly adduce rival theories in the course of their
reflexions about it. My discussion has shown the falsity of that
common assumption. ... The paradox, as we should expect and
desire, is innocent of any such theories: it presupposes only the
two harmless and common notions that there are instants, as well
as periods, of time; and that things move, if at all, at instants.
(Barnes (1982), p. 285)
As a purely exegetical remark, this may be above reproach, although
I am persuaded by Jonathan Lear's remarkable discussion of the
topic (Lear (1981), (1988)) that the best reconstruction of the
57
Robin Le Poidevin

argument gives prominence to the notion of the present moment, as

Barnes does not. I want to build on Lear's insight and offer a ver-
sion of the paradox that explicitly incorporates a theory of time,
thus (to my mind) increasing its interest and power. I cannot seri-
ously present the result as what Zeno originally intended, but we
should not expect, and need not insist, that reinvented wheels
should conform to the original blueprint.

II. First reconstruction: no motion in an instant

Although, as remarked above, our sources are not sympathetic ones,
they tend to converge in their accounts of the Arrow, which should
give us hope. The most argumentatively explicit account is that of
Simplicius:1
The flying missile occupies a space equal to itself at each instant,
and so during the whole time of its flight; what occupies a space
equal to itself at an instant is not in motion, since nothing is in
motion at an instant; but what is not in motion is at rest, since
everything is either in motion or at rest: therefore the flying mis-
sile, while it is in flight, is at rest during the whole time of its
flight. (Lee (1936), p. 53.)
This suggests the following reconstruction:
(1) If x moves throughout a period T, then x moves at each
instant of T.
For each instant t of T:
(2) x is either at rest or in motion at t.
(3) x occupies a space equal to itself at t.
(4) If x occupies a space equal to itself at t then x is not in motion
at t.
From (3) and (4):
(5) x is not in motion at t.
From (1) and (5):
(6) x does not move throughout T.
Finally, from (2) and (6):
(7) x is at rest throughout T.
' For sources and commentary, see Lee (1936), Barnes (1982) and Kirk,
Raven and Schofield (1983). For discussion, see Ross (1936), Owen (1957),
Vlastos (1966), Griinbaum (1967), Salmon (1970) and Sorabji (1983).
58
 Zeno's Arrow and the Significance of the Present

The crucial term of art here, of course, is 'instant'. For the premises
to be remotely plausible, we need to interpret this as an indivisible
point of time, not further resolvable into smaller items. If there can
be nothing smaller than an instant, then whatever takes place with-
in an instant cannot be differentiated into distinct states obtaining
in smaller moments. (I postpone until the next section the question
of whether these indivisibles have any duration.)
Having settled that point of interpretation, let us begin by indi-
cating some general lines of attack.
Aristotle's objection, which would threaten (2), that if one cannot
speak of something's being in motion at an instant, one cannot
properly speak of its being at rest at an instant either, can be con-
ceded without damaging the argument. (2) and (7) do not contribute
anything of value, since (6) is quite shocking enough on its own. So
we need to look at (1), (3) and (4).
Aristotle has another, more worrying objection: '[the conclusion]
follows from the assumption that time is composed of moments: if
this assumption is not granted, the conclusion will not follow.'
(Physics 239b30-3) The premise this observation is most relevant to
is (1), since one could defend (1) on the basis that there is nothing
to T other than the instants within it, and what is not true of any
instant cannot, be true of the period. But this rationale is dubious
whether or not we suppose time to be composed of instants. For to
infer a property of the whole from a property of the parts is com-
mit the fallacy of composition. Perhaps, then, we should present (1)
as an application of the following principle:
If x is F throughout T, x is F at every instant of T.
Now, clearly, for most properties, this principle is perfectly accept-
able: green, cubic, composed of copper, at 10 degrees C, etc. Nor is
this list confined to monadic properties. It seems just as acceptable
when applied to relational properties: being three yards from a heat
source, reflecting light, being an object of perception, etc. The prin-
ciple, then, is a plausible one, and seems to embody neither the falla-
cy of composition nor the assumption that Aristotle locates in Zeno's
argument. (It may be, however, that Aristotle's reasoning behind his
remark that time is not composed of moments is damaging in a dif-
ferent way to the argument; we return to this below.) (1), then, will
only be vulnerable if there is any good reason to doubt that the gen-
eral principle above applies to the property of being in motion.
What of (3)? A moving object occupies a succession of different
positions and so, in a sense, occupies (although not all at once) a
volume of space that is greater than its own—i.e. that whose
59
Robin Le Poidevin

boundaries are defined by the object's surface. Of course, the vol-

ume of an object will be defined as that which it occupies at an
instant, so it is trivially true that an object, whether moving or not,
occupies at an instant a space equal to its volume. (3), then, seems
secure. But what is inferred from this, namely that a moving object
is not moving at an instant, seems far from trivial. The most
obviously controversial premise, then, is (4).
However, before proceeding, we need to look at the suggestion
that the argument depends on a further, tacit, assumption about the
nature of instants.

III. Atomism: the hidden premise?

Why does Aristotle suppose the Arrow to depend on the premise

that time is composed of instants? As we saw in the previous sec-
tion, it is possible to defend (1) without recourse to such an assump-
tion, and indeed explicit appeal to it would invite the charge of fal-
lacy. Now it may be that Aristotle takes Zeno to be making a fairly
obvious blunder, but there is a more interesting possibility.
It is uncontroversial that the argument depends on realism about
instants. There must be such things as instants, not simply as logi-
cal fictions, but as independently existing entities, for there to be
truths about what is the case at those instants. But this realism may
imply something about the topological structure of time. Let us
imagine that the indivisible instants of time at which the Arrow
occupies a space just its own size have a small, but non-zero dura-
tion. Then periods of time would be composed of such 'time
atoms', and the number of such atoms in a period would determine
its length (assuming that each atom has the same duration). Time
atoms would be the fundamental unit from which periods were built
up. So a discrete topology for time is compatible with—indeed
implies—realism over instants. But now suppose instead that time
is not discrete but continuous, so that each period of time were
indefinitely divisible. What implications would this have for
instants? This is what I take to be Aristotle's answer to that ques-
tion: Although time is continuous, we avoid the consequence that
would appear to flow from this, namely that each period contains an
infinite number of instants, by the assumption of finitism. Nothing
is actually infinite (there is no infinitely large object, for example),
but only potentially infinite. So, for example, numbers are infinite
only in the sense that, however high a number one has counted to,
one can always count to a higher number. The process of counting
60
 Zeno's Arrow and the Significance of the Present

has no limit built into it. Similarly, a length is infinitely divisible

only in the sense that, however many divisions you have made, you
can always make more. The process of dividing has no limit. But the
divisions do not exist independently of one making them, so the
potentially infinite divisibility of a length does not imply the actual
existence of an infinite number of divisions within that length.2
And as with length, so with time: a period of time is infinitely divis-
ible, but only potentially so. Durationless instants do not exist inde-
pendently of their being marked in some way. This view of the
divisibility of space and time provides Aristotle with an answer to
two other Zenonian paradoxes of motion, the Dichotomy and the
Achilles. Both paradoxes start from the assumption of the continu-
ity of time and space, and derive the unpalatable conclusion that
motion involves achieving the impossible: the traversal of an infi-
nite number of sub-distances in a finite length of time. Aristotle
dissolves the paradoxes by denying that divisions have any inde-
pendent existence. Admittedly, Aristotle only explicitly applies this
to spatial divisions, but the same reasoning leads to a similar con-
clusion concerning temporal divisions.
This, then, is my conjecture: Aristotle, quite correctly, takes the
paradox to depend on realism about instants. But he takes this real-
ism to imply a discrete topology for time, which he expresses as the
view that time is composed of instants. So attributing to Zeno the
premise that time is composed of instants is simply an expression of
the fact that the Arrow depends on the real existence of instants.
Whether or not this reconstruction of Aristotle's thinking holds
water, it is certainly true that a number of commentators have
attributed an atomistic premise to Zeno. Lee (1936), for example,
takes it as obvious that this is an assumption (even if an implicit
one) without, however, giving his reasons for thinking so. There is
something attractive about the suggestion. For one thing, it presents
a pleasingly symmetrical picture of the overall dialectic of Zeno's
four famous paradoxes of motion. The Dichotomy and Achilles
show that motion is impossible if space and time are continuous; the
Arrow and Stadium show that motion is impossible if space and
time are discrete. Since motion is impossible without time, and time
must either have a discrete or continuous structure, 3 motion is
impossible.
2
See Physics, Book III, Chapters 6 and 7. Hussey (1983), pp. 14-18.
3
I ignore here, since it does not affect the argument, the possibility that
time might be merely dense: that is, isomorphic to the series of rational,
rather than real numbers. See Newton-Smith (1980) for a discussion of the
distinction between continuity, density and discreteness.
61
Robin Le Poidevin

Barnes calls this picture 'a neat fantasy', and I am inclined to

agree. What matters is that an instant is indivisible, thus permitting
no motion within it. But this does not conflict with a continuist
topology: it merely implies that if time is continuous, an instant must
be defined as having no duration whatsoever. For Barnes, then, the
argument need make no topological assumptions. This does not, of
course, provide an answer to the objection to instants raised by
Aristotle's finitism. If there is no actual infinite, how can there be an
infinite number of actual instants in any period? Perhaps finitism
can be reconciled with instants and the continuum, however, by
appeal to Aristotle's conception of the actual infinite as that which
exists all at once. A continuum of real spatial points would count as
an actual infinite by this definition, whereas a continuum of tempo-
ral points would not, as the points are successive and not simultane-
ous. (It has to be admitted, however, that this move somewhat weak-
ens the case for attributing to Aristotle the thought that Zeno's
Arrow depends on a discrete structure for time.)
Even if realism about instants can be reconciled with the contin-
uum, the Arrow can legitimately be challenged as follows: why
should we take realism about instants seriously, when we can say
everything we want to say about time using only the language of
intervals? Temporal order, metric and topology can all be expressed
in terms of intervals. And we can construct instants out of intervals
by treating them as the notional lower limit of nested intervals.4 So
why introduce instants when, as the Arrow shows, they lead to dif-
ficulties? Well, there is one instant at least whose existence is hard
to impugn: the present moment. What happens to Zeno's Arrow
when we build reference to the present into the reconstruction?

IV. Second reconstruction: no motion in the present

The most effective answer to our first reconstruction of the Arrow

is to produce an account of motion (and of change in general) that
is both plausible and which undermines at least one of the premis-
es. The obvious candidate for such an account defines change as fol-
lows: change consists in the instantiation at different times of
incompatible states of affairs. Thus, for example, just as heating
consists in an object's (or region's) being at different temperatures
at different times, so motion consists in an object's occupying
different places at different times. This approach to motion we will
call, after Russell (but also somewhat contentiously), the static
4
See Newton-Smith (1980), pp. 134-8.
62
 Zeno's Arrow and the Significance of the Present

account of motion. There are two ways in which one might take the
static account as undermining the Arrow. One is to take it as falsify-
ing (1). One could concede that there is no such thing as motion (or
change in general) at an instant, but merely the occupancy of a par-
ticular position (or state), and insist that motion is attributable to an
object only over a period of time. By analogy, an extended object
may be ten foot wide without it being true that all its parts are ten
foot wide. This is Russell's response to the Arrow.5 It is a rather
surprising response, however, since there is a kind of motion that
seemingly obliges us to talk of motion at an instant, namely accel-
eration. If an object accelerates continuously through a period, it
surely has a different velocity at each successive instant. One could,
perhaps, insist that this is merely a theoretical abstraction, but it
would have to go hand in hand with the view that instants them-
selves are theoretical abstractions. A much more plausible approach
takes our analysis of motion to undermine premise (4). There is no
need to deny that objects can move at an instant, but they do so only
in a derivative sense. An object is in motion at t iff it occupies dif-
ferent positions at times immediately preceding and/or immediately
succeeding t. What is true at an instant thus depends in part on what
is true at other times. So from the indisputable fact that an object
necessarily occupies a space just its own size at t, it does not follow
that an object cannot be in motion at t.
The game is not yet up, however, for we can provide another
reconstruction of the Arrow, one that brings out the limitations of
the static account of motion.
A feature of Aristotle's reconstruction of the Arrow, and of his
discussion of time in general, is the phrase 'in the now' (ev TO WV).
Jonathan Lear has suggested that, by interpreting nun as 'moment',
commentators like Owen, Vlastos and Barnes have overlooked the
significance of the nun: it is the present moment and not merely
some arbitrary instant. Lear's own reconstruction of the paradox
goes as follows:
(1) Anything that is occupying a space just its own size is at rest.
(2) A moving arrow, while it is moving, is moving in the present.
(3) But in the present the arrow is occupying a space just its own
size.
(4) Therefore in the present the arrow is at rest.
(5) Therefore a moving arrow, while it is moving, is at rest.
(Lear (1988), p. 84)
s
See Russell (1903), pp. 467-73. His definition of motion is given on p.
473. For a discussion of the Arrow, see pp. 350-2.
63
Robin Le Poidevin

Recall one objection to the first reconstruction: that we could

avoid talk of indivisible instants altogether in favour of intervals,
and in so doing make irrelevant the alleged fact that nothing moves
at an instant. But turning the focus of the argument onto the pre-
sent moment explains why the moment in question must be an
indivisible instant and not a period. For if the present were divis-
ible into different parts, some would be earlier than others, and so
not present. But every part of the present must itself be present,
which is to say, of course, that the present has no earlier and later
parts.
Another feature of Lear's reconstruction is that he talks, not of
what is happening at the present, but in the present (this being the
literal translation of 'ev TO WV'). This certainly makes the premises
somewhat more plausible, for we could happily concede that noth-
ing moves in (i.e. within the space of) the present. It is not immedi-
ately obvious how this helps, however, because we could still insist
that the crucial question is what is true of the arrow at an instant,
and, as Barnes points out,6 commonsense dictates that it is true to
say of the arrow that it is moving at an instant of time. It moves at
an instant by virtue of that instant being part of a period in which
the arrow is continuously in motion.
Nevertheless, the second reconstruction does represent a signifi-
cant advance on the first. The most powerful objection to the first
reconstruction is constituted by the static analysis of change: some-
thing moves in an instant by virtue of its position both at that
instant at other times. So talk of motion at an instant is derivative:
its truth depends on what is happening over a period of time. It is
this move that is challenged by the second reconstruction. For what
is true in the present should not be derivative, but fundamental. It is
the privileged status of the present that insulates present fact from
past and future fact. The static analysis of change makes expres-
sions like 'x moves in the present' temporally hybrid, turning what
purports to be a simple statement about the present into a complex
statement about past, present and future. But, we may imagine the
champions of the present arguing, 'x moves in the present' is a sim-
ple statement about the present, and should not be taken as ellipti-
cal for something else. Here is how Lear presents the moves:
To Zeno's incredulous question, 'So you think that an object can
be moving solely in virtue of positions it has occupied in the past
and will occupy in the future?', one would simply answer 'yes.'
This would be the response of someone who did not wish to
6
See Barnes (1982), pp. 280-1.
64
 Zeno's Arrow and the Significance of the Present

incorporate the notion of a present duration into his scientific

theory of time. (Lear (1988), p. 90)
I think Lear is quite right, by which I mean I think he has con-
structed a much more effective argument than one which talks
merely of instants. I also think he is right that the Arrow is most
problematic for those who accord a special status to the present, and
that one effective response (although one which requires a revision
to our ordinary view of time and change) is to dethrone the present
and treat it as a merely perspectival feature of our temporal descrip-
tions. But he moves rather too fast. We need a more explicit account
of what it is to treat the present as somehow special, and we also
need to be clear why this should imply anything about the nature of
change and motion. Before embarking on this, however, we should
consider whether there is a plausible alternative to the static account
of motion, implied in the suggestion that 'x moves in the present' is
a simple assertion about what is presently the case.

V. The dynamic account of motion

The static account of motion, recall, goes as follows:
An object moves at a time by virtue of its position at that time and
its position(s) at other times.
The contradictory of this should be something like the following:
The dynamic account of motion: an object's motion at a time is
independent of the object's position at other times.
This is not, however, a single account, but the common denomina-
tor of a number of accounts. For example, the statement above
leaves it an open matter whether an object's motion at a time is an
intrinsic property of that object or a relation between that object
and simultaneously existing objects or places. It is also an open
question whether the account arises from a demotion of instants to
a logical construction from intervals, as canvassed in §111.
Introducing greater specificity and detail into the dynamic account
would produce the following variants:
(i) It is an intrinsic property of an object that it is in motion at
a particular time, the property in question being a disposition
of the object to be elsewhere than the place it is.
(ii) Motion at an instant consists in the object both being, and
not being, in the place where it is.
65
Robin Le Poidevin

(iii) Events, including those involving motion, are primitive, not

decomposable into series of states. Similarly, intervals are
primitive, not decomposable into series of instants. So talk of
'motion at a time' must always be interpreted as motion in an
arbitrarily small interval.
Account (i) falls victim, I think, to the following counterexample: a
stationary object is struck at t by a rapidly moving one, as a result
of which (causation being successive and never simultaneous) it
begins to move after t. Though the object is disposed to be elsewhere
at t, as a result of its being subjected to a force at that moment, we
would count t as the last moment of rest, rather than the first
moment of motion, (i) would also conflict with the (more con-
tentious) assertion that constant motion is always relative to some
other object or objects.
Account (ii) is, most commentators would admit, very strange.
Priest (1987), who offers an imaginative and sympathetic recon-
struction of it, describes it as the 'Hegelean' account of motion, on
the basis of the following admittedly very explicit remark of
Hegel's:
...motion itself is contradiction's immediate existence. Something
moves not because at one moment of time it is here and at
another there, but because at one and the same moment it is here
and not here. (Hegel (1840), quoted by Priest (1987), p. 219)
To take it seriously is, in effect, to embrace dialetheism, the view that
there are true contradictions, and that contradictory states of affairs
can actually obtain.7 Even the dialetheist is not going to countenance
all contradictions, however, but only those for which there are com-
pelling reasons. Since the arguments of this paper tacitly employ the
Law of Non-Contradiction, which dialetheism rejects, however, I
cannot without hypocrisy leave logical space for the Hegelean
account. No doubt it deserves proper consideration, but here I will
simply point out that it is contradictory, and hope that by abandon-
ing it on those grounds I have not lost all my readers.
We are left, then, with (iii). It certainly undermines the first
reconstruction of the Arrow, for reasons already rehearsed. Does it
undermine the second? Let us reserve judgement for the time being,
and return to Lear's contention that the key concept in an under-
standing of the Arrow is that of the present. Now there is more than
one way to take the present seriously, the most obvious being to take
7
For an extended exposition and defence of dialetheism, see Priest
(1987).
66
 Zeno's Arrow and the Significance of the Present

it as the temporal location of all that is real. So let us now consider

its effect on Zeno's paradox.

VI. The consequences of presentism

The word 'presentism' is often used to describe the view that only
what is present is real (elsewhere (1991) I used the term 'temporal
solipsism', but here I follow convention). I take this to imply the
following theses:
(a) Where the domain of quantification is concrete objects, the
existential quantifier ranges only over those objects that are
located in the present moment.
(b) The truth-makers of past- and future-tensed token sentences
are present facts.
And, as a consequence of (a) and (b):
(c) Primitive relations obtain only between contemporaneous
objects.
One way of expressing presentism is to say that all truths, funda-
mentally, are present truths. What was and will be the case obtains
only by virtue of what is now the case. This is a much more sub-
stantial position than the trivial thesis that any propositions of the
form
It was the case that p
and
It will be the case that p
are equivalent to, respectively,
It is now the case that (it was the case that p)
It is now the case that (it will be the case that p)
Presentism is essentially an ontological thesis, restricting reality to a
single time. It is hard to think of a way in which the present moment
could be taken more seriously. If Lear is right, then, about the
importance of the present moment to the Arrow, presentism should
have implications for the treatment of motion. In particular, we
should expect that 'x moves in the present' cannot, on a presentist
reading, be treated as elliptical for a statement about past and future
positions.
However, at first sight, presentism seems rather to favour the
67
Robin Le Poidevin

static analysis, insofar as it conflicts with the rival to that analysis.

According to the version of the dynamic account which we were left
with at the end of the last section, an object's being in motion is a
primitive event, not further analysable in terms of objects, proper-
ties and times. Now for these primitive events to exist, on the pre-
sentist reading, they must be capable of existing in the present. But
events, being changes, are not instantaneous items: they take up
time. So, at best, what exists in the present are parts of events. The
idea of events having parts that are not themselves events, however,
conflicts with the primitive status of events. To the question, what
are these parts? the obvious answer seems to be: instantaneous states
of an object. Presentism is therefore incompatible with primitivism
about events. Thus we come back to the static analysis.
So why should we suppose there to be any conflict between pre-
sentism and the static analysis? For the presentist, all fact is present
fact. This implies, then, that x's moving in the present is simply a
present fact, or collection of present facts. This is at least formally
consistent with the static analysis. The presentist can allow that x
was in a different position from the one it now occupies, but has to
insist that this is made true by present fact. So one set of present
facts makes true 'x was at s/, another set makes true 'x is at s2', and
yet another set 'x will be at s3'. Thus present fact can, in principle,
make it true that x is moving, even when we understand motion in
terms of the static analysis.
It is clear that we will have to look a little deeper if we are to dis-
cern the kind of difficulty Lear was alluding to. The problem, I
think, has to do with the extent to which the present is capable of
making determinate past states of affairs. Firstly, anything less than
a fully deterministic universe will leave some propositions concern-
ing the past without a determinate truth-value (this feature is pecu-
liar to presentism, of course). Secondly, it is not even clear that the
presentist is in a position to build determinism into their world view.
Let us look at these points in turn. What presentism requires, in
order to guarantee the truth of propositions concerning the past
location of objects, is the classical determinist picture, expressible as
follows:
Given the conjunction of all laws, for any two times, t and t', the
total state of the universe at t is logically compossible with only
one total state at t'.
This particular formulation is not, however, essential to the classi-
cal position. We could, for example, have avoided explicit reference
to laws, treated determinism as a property of theories rather than
68
 Zeno's Arrow and the Significance of the Present

the universe, and replaced reference to total states of the universe

by reference to models of those theories, as follows:
The conjunction of all true theories is such that if any two
models of that conjunction are isomorphic at any point, they are
isomorphic at all other points.8
The appeal to truth here is essential. Even if determinism is a
feature of theories, the fact (if it is a fact) that the conjunction of all
true theories is deterministic will depend on a feature of the world.
Any non-realist interpretation of theories, or doubts whether there
could be such a thing as the conjunction of all true theories,
represent a departure from the classical picture.
Both formulations above, note, are time-symmetric: the state at
one time determines both later and earlier states. Some statements
of classical determinism are not time-symmetric, and make only the
future depend on the present, but it is clear that the presentist
requires the past to be determined too.
Even with determinism (of the relevant sort) in place, however,
problems remain, which brings us to the second point. Since pre-
sentism confines reality to a single, durationless point, it cannot
build in to the description of the total state of the universe at a given
time reference to motion, for this imports states at times other than
the present. All that is available to the presentist is the position of
objects, their various states, and the forces acting upon those
objects. Is this enough to determine earlier and later positions?
Consider the following two cases:
(i) The substantivalist conception of space as an entity existing
independently of its contents is correct. Consequently, there
is such a thing as absolute motion (this being simply motion
relative to space itself). Now consider a universe of objects in
absolute, but not relative, unidirectional and non-accelerating
motion. Can the position at any one time of those objects
determine earlier and later positions? No.
(ii) The substantivalist conception of space is not correct, and all
motion is relative to other contents of space. Now consider a
(relatively simple) universe in which the forces on certain objects
at a particular moment cancel each other out, so there is no
resultant force on those objects in any one direction. Can the
state of the objects at that moment determine their earlier and
later positions? No.
8
See Butterfield (1998) for different formulations of determinism and
discussion of associated difficulties.
69
Robin Le Poidevin

It seems, then, that we can say, first, that classical determinism can-
not accommodate absolute motion, if it is built on presentist
assumptions;9 and secondly, that at least in a range of cases, presen-
tism cannot even account for the relative motion of objects.
Consequently, the presentist can only adopt the static analysis of
motion (and change in general) on pain of an unacceptable level of
indeterminacy. On the other hand, since we have already ruled out
treating change in terms of primitive events, it is not clear what
other account is now available to the presentist.
Our third and final reconstruction of the argument can be
presented as follows:
(A) The dynamic account of motion: an object's motion at a
time is an irreducible property, independent of the object's
state at other times.
(B) The static account of motion: an object moves at a time by
virtue of its position at that time and at other times.
(1) If motion is possible, then either (A) or (B) is the correct
account of it.
(2) If presentism is true, (A) is false.
(3) If presentism is true, (B) is false.
From (1), (2) and (3):
(4) If presentism is true, motion is impossible.
What further conclusion we draw from (4) I leave to the reader.10
I have tried to provide some more detailed argument for Lear's
suggestion that what we have called the static account of motion is
only available to 'someone who did not wish to incorporate the
notion of a present duration into his scientific theory of time.'
There is, I have argued, a conflict between presentism and the sta-
tic account. But is the choice of positions limited to presentism on
the one hand and an elimination of the present in favour of a pure-
9
Priest (1987), p. 217, notes a tension between classical determinism and
the static account. That tension, however, could be resolved by a small
revision to the classical account, as follows: states of the universe at arbi-
trarily (but non-zero) intervals determine states at other times. This move,
of course, is blocked by presentism.
10
I cannot resist adding another twist, however. Consider the following
paradox, as formulated by Owen in his discussion of Aristotle's conception
of time: 'Only the present is real, yet the present is never a stretch of time.
... Moreover time is a function of change ... but nothing can be changing
at a present moment. How can time be real?' (Owen (1976), p. 309: page
reference to the reprinted version.)
70
 Zeno's Arrow and the Significance of the Present

ly tenseless conception of time on the other? Is there no room for a

view that takes tense to be part of the world, rather than simply our
representations of it, and yet which does not restrict reality to the
present? For reasons I have articulated elsewhere ((1991), Chapter
2), I believe the answer to this question to be 'no'. Treating tense as
real can only lead to contradiction unless one assumes presentism.
Our final reconstruction of the Arrow would horrify Barnes (and
perhaps Zeno, too). But then an Arrow purged of any but the most
minimal theoretical content is vulnerable, as we have seen, to rather
obvious objections. To strengthen the Arrow, we need to engage
with theory. If one of history's most ancient paradoxes is to survive
into the new age, that's the deal.

References

Barnes, Jonathan 1982. The Presocratic Philosophers, revised edition,

London: Routledge and Kegan Paul.
Butterfield, Jeremy 1998. 'Determinism and Indeterminism', in Edward
Craig (ed.) The Routledge Encyclopedia of Philosophy, London:
Routledge, pp. 33-9.
Griinbaum, Adolf 1967. Modern Science and Zeno's Paradoxes,
Connecticut: Wesleyan University Press.
Hegel, G. W. F. 1840. Lectures on the History of Philosophy, trans. E. S.
Haldane, London: Kegan Paul.
Hussey, Edward 1983. Aristotle's Physics, Books III and IV, Oxford:
Clarendon Press.
Kirk, G. S., Raven, J. E., and Schofield, M. 1983. The Presocratic
Philosophers: A Critical History with a Selection of Texts, 2 nd edition,
Cambridge: Cambridge University Press.
Lear, Jonathan 1981. 'A Note on Zeno's Arrow', Phronesis, 26, pp. 91-104
(1988) Aristotle: The Desire to Understand, Cambridge: Cambridge
University Press.
Lee, H. D. P. 1936. Zeno of Elea, Cambridge: Cambridge University Press
Le Poidevin, Robin 1991. Change, Cause and Contradiction, London:
Macmillan.
Newton-Smith, W. H. 1980. The Structure of Time, London: Routledge
and Kegan Paul.
Owen, G. E. L. 1957. 'Zeno and the Mathematicians', Proceedings of the
Aristotelian Society, 58, pp. 199-222; reprinted in Owen (1986), pp.
45-61.
1976. 'Aristotle on Time', in P. Machamer and R. Turnbull (eds),
Motion and Time, Space and Matter, Columbus: Ohio State University
Press, pp. 3-27; reprinted in Owen (1986), pp. 295-314.
1986. Logic, Science and Dialectic, ed. Martha Nussbaum, London:
Duckworth.
71
Robin Le Poidevin

Priest, Graham 1987. In Contradiction: A Study of the Transconsistent,

Dordrecht: Nijhoff.
Ross, W. D, 1936. Aristotle's Physics, Oxford Clarendon Press.
Russell, Bertrand 1903. The Principles of Mathematics, Cambridge:
Cambridge University Press.
Salmon, Wesley 1970. Zeno's Paradoxes, Indianapolis: Bobbs-Merrill.
Sorabji, Richard 1983. Time, Creation and the Continuum, London:
Duckworth.
Vlastos, G. 1966. 'A Note on Zeno's Arrow', Phronesis, 11, pp. 3-18.

72
Presentism, Ontology and Temporal
Experience
L. NATHAN O A K L A N D E R

In a recent article, 'Tensed Time and Our Differential Experience

of the Past and Future,' William Lane Craig (1999a) attempts to
resuscitate A. N. Prior's (1959) 'Thank Goodness' argument against
the B-theory by combining it with Plantinga's (1983) views about
basic beliefs. In essence Craig's view is that since there is a univer-
sal experience and belief in the objectivity of tense and the reality
of becoming, (that he identifies with 'the presentist metaphysic')
'this belief constitutes an intrinsic defeater-defeater which over-
whelms the objections brought against it.' (1999a, 519) An intrinsic
defeater-defeater is a belief that enjoys such warrant for us that it
simply overwhelms the defeaters brought against it without specif-
ically rebutting or undercutting them. Thus, Craig claims that an
effete philosophical argument like McTaggart's paradox is nothing
more than 'an engaging and recalcitrant brain teaser whose conclu-
sion nobody really takes seriously.' (1999a, 532) It is difficult to rec-
oncile this statement with Craig's own writings elsewhere. For
Craig has vigorously argued in at least two other articles that
'hybrid A-B theorists like McCall, Schlesinger, and Smith [who
give ontological status to both A-properties and B-relations] are in
deep trouble' (1998, 127) since they are all effectively refuted by
McTaggart's Paradox (cf. Craig 1997). It is not Craig's inconsisten-
cy regarding the significance of McTaggart conundrum that I want
to draw attention to, however. Rather I wish to raise a different
issue.
Presentists such as Prior and Craig (and A-theorist's generally)
maintain that one motivation, and for some the primary motivation
for adopting presentism is 'the desire to do justice to the feeling that
what's in the past is over and done with, and that what's in the
future only matters because it will eventually be present.'
(Zimmerman 1998, 212; my emphasis). As Craig puts it, '[O]nly on
the A-theory; with its ontological distinctions between past, pre-
sent, and future can differential attitudes toward events ... be ratio-
nally justified.' (1999a, 530) He claims that one is rationally justi-
fied in feeling relief concerning the cessation of a painful experience
because
73
L. Nathan Oaklander

[O]n a presentist metaphysic the experience was once real and

now no longer is. ... (Analogously, dread about some future
painful event is appropriate because, although not yet real, it soon
will be, ...). (1999a, 521-22)
In these passages the phases 'eventually will be present,' 'once real'
and 'it soon will be' seem to imply the existence of temporal rela-
tions. For, 'once' in this context means, 'at some earlier time,' 'soon'
means 'at a relatively short time later,' and 'eventually' means 'at a
relatively distant later time.' Thus, to explain our differing attitudes
of dread and relief it would appear that in addition to the existence
of the present time (and present experiences), earlier and later times
(and earlier and later experiences) must be real too. As Craig him-
self says, 'When I feel relief, what I am relieved about can be
analysed as a complex fact that it is now and that the relevant event
is earlier than now.' (1999a, 523; my emphasis)
With this background the main question I wish to explore in this
paper can be stated as follows: Can an ontological analysis that spec-
ifies the constituents of the complex fact that it is now and the rele-
vant event is earlier or later than now be given that is consistent with
presentism? Prima facie the answer to this question is 'no' because
in order for there to be a temporal relation between two events there
must be the two events that stand in that relation. Indeed, this is a
truism that is even accepted by some presentists. Consider, for
example, the following statement by John Bigelow, whose views I
discuss in detail below:
It is; I maintain, an a priori truth that a two-place relation can
only be manifested when it holds between two things, and in
order for this to be so there must be two things which stand in the
relation. And in saying 'there must be' two things which stand in
the relation, one is really asserting that 'there must exist' two
things—one is committed to the existence of those things. The
principle of the existence entailment of relations is an a priori
truth. (1996, 39)
Since relief involves a temporal relation between a present belief or
feeling, and something else that is not present (what I believe or am
relieved about), it follows that some things/events/times exist that
are not present. To avoid this conclusion presentists must provide
an ontological reduction of temporal relations, but it is not clear that
they have the resources available to accomplish that task. Thus, we
are led to ask, what is the truth-maker or, to use Robin Le
Poidevin's turn of phrase, 'that bit of reality' (1999b, 149) that is the

74
 Presentism, Ontology and Temporal Experience

ontological ground of the complex fact that one event (or time) is
earlier or later than the present event or time? If presentists cannot
give an adequate answer to that question then far from rationally
justifying our different attitudes toward earlier and later events,
they render those attitudes mysterious.
The importance of giving an adequate ontological assay of tem-
poral relations is highlighted by another concern. Presentism has
been claimed to border on the trivial or the absurd. It is trivial if
it is taken to assert, 'The only things that exist now (i.e. at present)
are those that exist at present.' (Zimmerman 1998, 209) And it is
absurd if it is taken to mean what David Lewis takes it to mean,
namely, that for the presentist nothing changes because it rejects
persistence altogether by maintaining that there are no past or
future times, but only this present moment. (1986, 202-3) In
response to the triviality objection a presentist may claim that
their view does not assert the tautology that only the present exists
at present, but that only the present exists simpliciter or just plain
exists. To exist simpliciter does not mean to exist now, or to exist
at time t, but simply to exist (or to happen) without temporal
qualification.1 But if only the present moment exists without
qualification, then in order to generate change the presentist must
introduce tensed propositions that change their truth value at
different times depending on what exists simpliciter at those times
as time passes. Of course, if a tensed proposition changes its
truth-value at different times, then there must be those times at
which the proposition changes, and in order to be genuine times
they must be members of temporal series whose generating
relation is earlier/later than. But if there exists only one time, the
present time, then how can any time be earlier or later than
another and how can a proposition or anything else change from
one time to another}
How, in other words, can it be true, as it obviously in some sense
is, that

(PC) There are (at least) two different times, one at which I am
bent, another at which I am straight?
1
It seems, therefore, that the presentist is committed to Tooley's (1997)
distinction between what actual simpliciter and what is actual as of a time.
For if what exists simpliciter is all that exists, and if only the present
moment exists simpliciter, then Lewis is right and we have no past or
future. For a critique of the conjunction of the notions of actual simpliciter
and actual as of a time, see Le Poidevin (1998b, 2001), Oaklander (1999a),
Smith (1999a) and Mellor (1998).
75
L. Nathan Oaklander

In response to this question a presentist may offer a paraphrase that

captures what is meant by (PC) but does not involve direct reference
to non-present times. Thus, for example, Dean Zimmerman says
that (PC) can be taken as a tenseless statement expressing a dis-
junction of tensed propositions:
Either I was bent and would become or had previously been
straight, or I was straight and would become or had previously
been bent, or I will be bent and will have been or be about to
become straight, or I will be straight and will have been or be
about to become bent. Surely this tensed disjunction is true if
(PC) is true; furthermore, it contains no mention of anything like
a non-present time. So given the presentist's desire to avoid onto-
logical commitment to non-present times, this tensed statement
provides a perfectly sensible paraphrase of my conviction that I
can persist through change of shape. (1998, 215; emphasis added)
I shall avoid the question of whether Zimmerman's paraphrase cap-
tures what is meant by (PC)2 and concentrate on the question, 'Does
it avoid ontological commitment to non-present times?' It certainly
does not avoid commitment to temporal relations and, on the face of
it, temporal relations presuppose the existence of earlier and later
(i.e., non-present) times. Included in his analysis of (PC) are the
phrases 'had previously been straight,' and 'about to become bent.'
The former phrase has the sense of 'was at some earlier time
straight' and the latter means, 'at a short time later becomes (or will
become) bent.' Furthermore, the concept of temporal becoming, if
it is to be the ground of temporal change, that is, change in a given
direction, must account for my becoming straight before becoming
bent or vice versa. Thus, the presentist must provide an account of
temporal relations, and to provide such an account is to specify what
there is in the world, independently of minds, that is the truth
maker of judgments asserting that two entities stand in a temporal
relation. If an adequate account is not forthcoming then not only
are our different attitudes toward earlier and later events left unex-
plained, but Lewis's objection, that for the presentist nothing
changes, is vindicated. In what follows I shall consider several
recent versions of presentism that attempt to respond to the chal-
lenge of providing an adequate ontological ground of temporal rela-
tions and argue that none of them is successful.
Each of the philosophers I shall discuss, William Craig, John
Bigelow and Robert Ludlow, all avowed presentists, acknowledge
2
And I shall also avoid diagnosing why Zimmerman characterizes his
paraphrase of (PC) as both a 'tenseless statement' and a 'tensed statement.'
76
 Presentism, Ontology and Temporal Experience

their debt to Prior, but for one reason or another find his particular
explication of presentism wanting. Prior's views have also recently
received extensive critical discussion by other A-theorists such as
Craig (1997, 27-29), Smith (1986; 1993, 158-169; 1999b, 248-249,
2002), and Tooley (1997, 165-70, 232-8), as well as by B-theorists
such as Le Poidevin (1991, 36-57), Mellor (1998, 70-81) and
Oaklander (1984, 90-104). For that reason there will be only an
incidental discussion of Prior's views on time in this paper.
Let me then first consider Craig's various versions of presentism
to see if any of them can make sense of the notion of events stand-
ing in temporal relations. According to Craig,
[T]he A-theorist denies the very reality of past and future events,
but he does not deny that some events are objectively past or oth-
ers future—he just parses such statements to mean that 'It was
the case that e5 exists' and 'It will be the case that e2 exists' are
true. (1999a, 534)
On the face of it, to deny the reality of past and future events, and
to affirm that some events are objectively past and future is an
explicit contradiction. Thus, to avoid that contradiction we need to
ask of Craig's paraphrase of statements about past and future
events, what is it that makes them true? To that he responds:
[Corresponding to past/future-tense propositions are tensed
facts or states of affairs that presently obtain, e.g., its being the
case that e will occur [or it being the case that e did occur]. But e
[or e'] itself is not presently existent or real. Though equally
unreal, past and future events are properly regarded differential-
ly by us due to the direction of time. ... [O]n the A-theory the
impossibility of backward becoming entails the propriety of dif-
ferential attitudes toward earlier and later events. (1999a, 534)
One problem with this analysis is that if neither ex or e2 nor the
properties of pastness and futurity exist, then what could be the
foundation of the difference between the tensed facts that e2 will
occur and e^ did occur? And what could be the truth-maker for the
state of affairs that ej is earlier than e2?
Perhaps Craig would attempt to specify a tensed fact by appeal-
ing to the properties or a description that one associates with the
purportedly past or future event. He could then claim that the dif-
ference between a past and future tensed fact is that in one case the
defining properties of e2 will be exemplified, presently obtains
whereas in the other case the defining properties of ej once was
exemplified, presently obtains. However, I don't think this gambit
77
L. Nathan Oaklander

works. For unless one explains (or specifies) the ontological corre-
late of 'will be exemplified' and 'was exemplified,' there really is no
basis for the difference in the temporal location of either of these
facts. In other words, if there is no grounding of the past and future
tenses in the complex fact that e2 will be exemplified and ej was
exemplified, then there is no basis for determining whether ej is
exemplified before e2 is exemplified or vice versa.
Furthermore, I do not see how, as Craig says, the direction of
time can be the basis of the difference between past and future tense
states of affairs. On the A-theory, the direction of time is based on
the changing truth of past and future tense propositions. That is, on
the A-theory the direction of time is grounded in the fact that an
event is first future, and then present and then past rather than the
other way around. As Prior puts it,
We all know what it is to wait for something.... What we're wait-
ing for begins by being future; it hasn't yet come to pass. Then a
time comes when it does come to pass—when it's present, and we're
aware of its presentness, and there's no mistaking it. And then it's
past.... (1996, 50; my emphasis)
In other words, on the A-theory the direction of time is grounded in
an event/thing/time being future before it is present and being pre-
sent before it is past.3 Thus, the difference between past and future
events (or times), and the temporal relations that hold between them,
cannot be based on the direction of time because the direction of
time is based on the distinction between past and future tensed facts
and their temporal relations to the present. Furthermore, Craig's
claim that 'on the A-theory the impossibility of backward becoming
entails the propriety of differential attitudes toward earlier and later
events' (1999a, 534) is question begging. Without a prior account of
the ontological difference between past and future tensed facts, and
a grounding of temporal relations, Craig is not entitled to assume
that backward becoming is impossible.
3
Prior seems to be making essentially the same point when he says: 'I
believe that what we see as a progress of events is a progress of events, a
coming to pass of one thing after another, and not just a timeless tapestry
with everything stuck there for good and all.' (1986, 104) I should note,
parenthetically, that Prior's claim that what we see as a progress of events
(one thing occurring after another) is, on the B-theory, 'just a timeless
tapestry with everything stuck there for good and all,' is a common mis-
characterization that depends on viewing time from a God-like third per-
son point of view and not the subjective first-person temporal point of
view that B-theorists believe in.
78
 Presentism, Ontology and Temporal Experience

Nor can Craig appeal to causation to ground the direction of time

and the impossibility of backward becoming since the A-theorist
has just as much problem with causal relations as he or she does with
temporal relations because some present events are caused by non-
present events. After all, why believe, for example, that my presently
having lines around my eyes, or my presently having relatively
little hair on the top of head, is evidence that I have aged over the
past twenty years unless present evidence is causally related to the
past which thereby must exist? In other words, if only the present
exists how could the past be causally related to it?
In his article, 'Is Presentness a Property?' Craig realizes that the
presentist needs to somehow ground the relational aspect of time
since he says,
on a presentist ontology past and future events/things/times are
not real or existent and, hence, do not exemplify properties like
pastness and futurity ... The A-theorist thus agrees with the B-
theorist that pastness and futurity are relational predicates, but he
will differ in anchoring these relations in what is non-relationally
present. The construal of pastness and futurity as relational pred-
icates should not be taken to mean that these are relational prop-
erties inhering in events. Rather such ascriptions should be
parsed as asserting that the entity in question did or will exist.
(1997, 29; my emphasis)
Again, this move seems to me unavailing. To assert that ej did exist
and e2 will exist says nothing more than that ej is past and e2 is
future. Since, however, 'is past' and 'is future' do not name tempo-
ral monadic properties or relational properties of events or times,
Craig's paraphrase tells us nothing about the ontological difference
between past and future tensed facts or about how temporal
relations are to be 'anchored' in what is non-relationally present.
In adopting a presentist metaphysics Craig expresses his sympa-
thy with Prior's view that to be present is simply to exist. He recog-
nizes, however, that the treatment of 'was' and 'will be' and the past
and future tenses as 'tensed operators' analogous to the sentential
operator of negation, or the modal operator of possibility 'raises a
host of questions for ontology. [And it] makes one suspicious that
the ontological questions concerning tense ascriptions cannot be so
neatly circumvented as Prior hoped.' (1997, 31) This seems to me
to be quite true. Even if 'It was the case that ' and 'It will be the case
that' are tensed operators analogous to sentential operators like
negation (see Hinchliff, 1996), there remains the ontological ques-
tion concerning whether or not there are past and future tensed
79
L. Nathan Oaklander

facts. Although Russell allegedly caused a riot at Harvard when he

lectured about negative facts, Russell (1918) believed and argued
that negative facts exist. He is not alone, Gustav Bergmann (1964),
Reinhardt Grossmann (1992), Herbert Hochberg (1969) and others
have agreed and argued that there are conjunctive and general facts
as well.4 Hence, while treating the past and future tenses as opera-
tors does not imply that there are future and past tensed properties
or facts, it does not rule them out either. Nor can treating the past
and future tenses as primitive operators eo ipso avoid ontological
commitment to past and future individuals and A-properties. Since
even if the tenses are conceptually primitive one can still ask what
entities do those primitives stand for? I shall return to this point in
my discussion of Peter Ludlow.
In another paper 'McTaggart's Paradox and the Problem of
Temporary Intrinsics,' Craig (1998) attempts to explicate the ontol-
ogy of presentism 'by allowing tensed states of affairs to be con-
stituents of possible worlds. ... Tensed possible worlds which did,
do or will obtain are tensed actual worlds. ... A tensed actual world
at t is the world which obtains when t's being present obtains' (1998,
126), but when does t's being present obtain? Judging from his com-
ments it appears that t's being present obtains before t*'s being pre-
sent obtains (for any later t*), since Craig maintains that
The tensed history of any possible world W will be all the tensed
possible worlds constituted by the states of affairs entailed by W
and each successive t's being present in W. (1998, 126; emphasis
added)
Thus, to explain when possible worlds obtain he appeals to succes-
sion, but since the appeal to succession implies the reality of B-rela-
tions one of whose terms is non-present, this particular explication
does not answer the question of how a presentist ontology can con-
sistently account for the relational aspect of time. It simply assumes
that it can.
In his reply to my critique (1999b) Craig claims that his charac-
terization of presentism in terms of possible worlds did not mean to
'found [or ground] the objectivity of temporal becoming, but simply
to provide a language in which to formulate such notions.' (1999b,
320) If, however, the appeal to possible worlds being actual at times
that are present is not intended to explicate what makes relational
temporal statements true, then it does not address the fundamental
4
For an argument against negative facts that is also a precursor of one
version of the new tenseless theory of time see Oaklander and Miracchi
(1980).
80
 Presentism, Ontology and Temporal Experience

problem with the ontology of presentism. For it does not provide an

ontological explanation of how there can be objectively earlier and
later times if all that exists is present.
Finally, in another recent publication on time, 'The Extent of the
Present,' Craig maintains that the present is neither instantaneous
nor atomic, but is a pre-metrical notion that denotes some arbitrar-
ily selected finite interval. He continues,
Any temporal interval which is contextually taken to be the pre-
sent interval is susceptible of being conceptually divided into
shorter temporal phases which will be past, present, and future,
respectively. ... The present minute can thus be analysed into a
past phase composed of seconds earlier than the present second,
a present phase which is the present second, and a future phase
composed of the later seconds remaining in the minute. (2000a,
179)
Thus, to avoid Augustine's problem of a durationless present, Craig
claims that an interval can be present as a whole even if it is com-
posed of parts some of which are past and some of which are future.
It does seem to me that one advantage of this account is that it
looks like it has a chance of grounding the existence of past and
future tensed facts, as well as temporal relations between past, present
and future individuals in terms of what 'presently' exists.3 The dis-
advantage is that Craig's latest account is not compatible with pre-
sentism, as his comments on the 'present' make clear. He says that,
... an interval may be present simpliciter even though we can
divide it into sub-intervals which are not every one present.
Thus, the present minute is qua minute present simpliciter, but if
we divide it into seconds, then only one second is qua second pre-
sent simpliciter. If any sub-interval of an interval is present, then
the whole interval is as such present (2000a, 184).
If the present exists simpliciter, it would seem to be composed of
phases some of which exist simpliciter. And if past and future
phases are temporally related to the sub-interval that is present, and
5
I say that Craig's view looks like it has a chance to succeed, but it can-
not in fact succeed even if he countenances the full range of A-properties
and B-relations. For the most elaborate and carefully crafted A/B ontology
of tensed time is to be found in Smith. (1993 and 1994) However, in
Oaklander (1996) I argue that Smith's version of the A-theory cannot
account for events having their A-properties successively. Smith's view
also receives a trenchant criticism in Nerlich (1998). I should note, how-
ever, that in his paper for this volume Smith (2002b) has modified his
views in the light of criticism.
81
L. Nathan Oaklander

the entire interval exists because it is as such present, then it seems

to me that we have a tenselessly existing series whose sub-intervals
successively become present as time flows. This is the A-B ontology
that Craig claims is 'incoherent.' (2000a, 165) Whether it is coher-
ent or not, it is certainly not presentism.
For these reasons I do not think that Craig's various explications
of presentism have met the challenge to give an adequate ontologi-
cal assay of the complex fact that, say, an unpleasant visit to the den-
tist is earlier than a present memory of it, that is consistent with
presentism. It should be noted, however, that Craig has several
forthcoming articles and books on time in which he refines and
defends presentism. Perhaps in one of those writings a more plau-
sible version of presentism is to be found.6
I shall turn next to two other recent attempts (the first by John
Bigelow and the second by Peter Ludlow) to deal with the challenge
to presentism of what Bigelow aptly calls 'the argument from rela-
tions.' Bigelow attempts to ground truths about past and future
individuals and the temporal relations between them by means of
properties that are presently exemplified. He says that in order to
meet the argument from relations,
[W]e do not need to suppose the existence of any past or future
things, only the possession by present things of properties and
accidents expressed using the past and future tenses. ... Present
things have present properties and these are the ontological
ground of the past, the future and the passage of time. (1996, 46)
These properties may include things like the property of being
burdened with a certain sort of past, or (as Leibniz put it) being
pregnant with a certain sort of future. (1996, 47)
On Bigelow's view, presently instantiated properties are the ground
of the difference between the past and the future, and of temporal
relations. No temporal relation ever in fact holds between things
that exist at different times, since at any given time the ground of
temporal relations are properties that are presently exemplified by
the world as a whole at that time. Thus, Bigelow says,
[O]ne of the things that exist is the whole world, the totality of
things that exist. The world can have properties and accidents,
just as its parts may have. It is a present property of the world
that it is a world in which Helen was abducted, and the Trojans
were conquered. (1996, 46)
6
Craig's books on time (2000b, 2000c, 2001) have appeared since the
completion of this paper.
82
 Presentism, Ontology and Temporal Experience

And in discussing causal relations that, like temporal relations, pur-

port to connect present events with events that are not present, he
says:
[T]he causal relation does not, in fact, ever hold between things
that exist at different times. At any given time the causal relation
holds between properties, perhaps between world properties,
each of which is present and is presently instantiated. (1996, 46)
But can presently exemplified world properties of the sort Bigelow
introduces account for the difference between the past and the
future without introducing past and future individuals? Can they
account for the relational aspect of time? And is the positing of such
properties consistent with presentism? I do not think any of these
questions can be answered affirmatively.
Bigelow says that there are presently exemplified properties that
are expressed by the past and future tenses, but he never makes clear
what it is exactly that makes these properties tensed. To bring the
issue into sharper focus suppose at any given time both Helen was
abducted and Helen will be abducted are presently exemplified.
(That is, suppose that Helen was abducted, rescued and will be
abducted again.) What is the difference in these presently exempli-
fied properties expressed by the past and future tenses? How does
the fact that the world exemplifies the one differ from the fact that
the world exemplifies the other? And finally, how do those proper-
ties or facts provide an ontological ground of temporal relations and
the passage of time? Unfortunately, Bigelow does not directly
answer these questions, and it does not seem to me that he can
unless he countenances the non-relational temporal properties of
pastness and futurity. However, positing the existence of pastness
and futurity as constituents of past and present tensed world prop-
erties is problematic. For if there is a full range of tensed properties
then presumably there are past and future individuals as well. To
see why note that if the world presently exemplifies the property of
being such that the birth of my first grandchild will be past, then
that event exists in the future when it exemplifies pastness. And if
the world exemplifies the property of being such that the birth of
my first child was future, then that event exists in the past when it
exemplifies futurity. Thus, if some events did exemplify futurity
and others will exemplify pastness, then there are non-present, i.e.,
past and future times/events/things that exemplify those properties.
For that reason, if Bigelow posits the existence of pastness and
futurity to account for what is expressed by the past and future
tenses, then he must jettison presentism.

83
L. Nathan Oaklander

On the other hand, if tensed properties don't exist, then the dif-
ference between what is expressed by past and future tensed propo-
sitions remains mysterious, and there is no basis for the passage of
time. For, if I understand Bigelow's position correctly, the passage
of time requires that a future tensed proposition is true (and thus
that a future tensed property is presently exemplified) before a past
tensed proposition is true (and thus that a past tensed property is
presently exemplified). If, however, pastness and futurity are not
constituents of the properties expressed by the past and future
tense, then there is no difference between past and future tensed
facts, and for that reason, there can be no basis for the passage of
time and the direction of becoming.
There is a further problem closely connected with the previous
one. It is not clear how Bigelow would handle the problem of order-
ing events (or things) that are both past or both future. In other
words, since Lincoln's assassination is past and Kennedy's assassi-
nation is past, then on Bigelow's view, the world presently exempli-
fies the properties of Kennedy was assassinated and Lincoln was
assassinated. What, then, is the ground of the fact that Kennedy
was assassinated after Lincoln was assassinated? The typical move
of introducing degrees of pastness or futurity is not open to a pre-
sentist who rejects the properties of pastness and futurity.
Moreover, introducing properties such as being past by a certain
degree would not be sufficient to order the properties, and hence the
terms that exemplify those properties, unless there was a temporal
relation between the properties, or unless the property itself was
complex that included a relation. That is, if we order the events by
means of the world property that say, Lincoln's assassination was
more past then Kennedy's assassination, we are clearly reintroduc-
ing temporal relations as basic entities back into the ontological
analysis of temporal facts. Finally, the attempt to order past events
by appealing to a property such as Kennedy was assassinated 27
years ago whereas Lincoln was assassinated 135 years ago is ques-
tionable since, once again, it is not at all clear that different tempo-
ral intervals from the present can be accounted for in this manner
without presupposing either temporal relations or temporal proper-
ties such as pastness and futurity. (For further discussion of this
issue see, Le Poidevin (1999a, 30-35) and Tooley (1997, 166-70.)
I have one final point concerning Bigelow's world properties
gambit. If one accepts the full range of A-properties, as I believe
Bigelow is committed to doing by introducing tensed properties
such as Helen was abducted and Bush will be elected, then, as I sug-
gested above, he must accept past and future events/thing/times in
84
 Presentism, Ontology and Temporal Experience

addition to the present events/things/times. Bigelow says things that

drive him perilously close to explicitly espousing such a view.
Consider the following passages:
The past no longer exists; yet there is a sense in which the past
can never be lost: the world will always be one with the property
of having once been thus and so. Likewise the future does not
exist yet; yet there is a sense in which the future will be what it
will be: the world has always been one with the property of being
a world which is going to be thus and so. At any given time, you
can grasp truths which transcend your present and describe the
world sub specie aeternitatis, from the standpoint of eternity.
(1996, 47)
And again,
From the standpoint of presentism the rival theory four-dimen-
sionalism, is just a partial picture of reality. It only acknowledges
the eternal truths about what was, is or will be. The things it
acknowledges to be true are indeed true as far as they go, but they
are all accommodated as an abstracted, logical consequence of
what is real according to the presentist. (1996, 48)
Bigelow is correct in saying that his view implies that four-dimen-
sionalism is a partial picture of reality. It is, however, not an
abstracted logical consequence, but a metaphysical commitment
resulting from his acceptance of past and future tensed properties.
If we add to this partial picture of reality the view that Bigelow also
wishes to endorse, namely, that 'the true Present, the world, is not
identical with eternity. The world is rather a changing ground for
unchanging truths' (1996, 48), then we get a version of the moving
NOW A-theory that, whether immune to dialectical difficulties or
not, is certainly incompatible with presentism.
I want to conclude my critique of presentism with a brief discus-
sion of Robert Ludlow's version of that doctrine as found in his
recent book, Semantics, Tense and Time: An Essay in the Metaphysics
of Natural Language (1999). Like virtually all contemporary pre-
sentists, Ludlow's work takes its cue from Prior's writings on time
and tries to answer two difficulties that Prior's logical analysis of
tense sought to resolve. The first is McTaggart's paradox and the
second is the problem of temporal anaphora. The elegance of
Ludlow's own view of time is that it allegedly provides a common
solution to both problems. The weakness, I submit, is that it fails in
both attempts.
The problem of temporal anaphora is that of specifying, within a
85
L. Nathan Oaklander

presentist framework, the semantic value of implicit or explicit tem-

poral anaphors such as 'then' that seem to refer to non-present
times and events. For example, 'Sam addressed Bill. Bill didn't
respond then.' McTaggart's problem or paradox rests on the claim
that every event is past, present and future, which is absurd. Of
course, it has seemed to many that this absurdity need not trouble
us because we can say either that
(1) Every event is past at one time, present at another time and
future at a still different time.
Or, that
(2) Every event is future before it is present and present before it is
past.
Or, as Prior has said, that
(3) Every event either is future and will be present and past, or has
been future and is present and will be past, or has been future
and present and is past. (Prior 1967, 5—6)
Each of these ways of avoiding the original absurdity results in a
statement that removes any explicit contradiction in temporal attri-
butions, but whether they involve anything more than a verbal
solution to an ontological problem is debatable. Since the first alter-
native presupposes the existence of past and future times and the
second presupposes the existence of temporal relations, both are an
anathema to the presentist. Thus, the third option is clearly the one
a presentist must take, but it gives rise to the following question:
How is one to analyse the tenses? More specifically, what are the
truth-makers of each of the conjuncts in each disjunct? Is there
some analysis of the tenses that avoids collapsing the third alterna-
tive into a variation of one of the first two? For example, can one
analyse 'event e has been present' in such a way that it avoids an
ontological commitment to either a past time at which e is present or
to e's occurring earlier than a present moment?
Ludlow thinks that the A-theory has presentist resources avail-
able to answer these questions since he believes that we can
treat the standard B-theory predicates 'before and 'after' as com-
posed out of more basic A-series relations. The idea here would
be that a sentence like [I ate before I left the house] would have a
logical form in which 'before' is treated as composed of a past-
tense morpheme and a simple when-clause. (1999, 126)
I do not need to go into details here. The overall point is that we can
86
 Presentism, Ontology and Temporal Experience

eliminate reference to times and temporal relations between them

by giving the semantics for 'before' and 'after' in terms of the pred-
icates 'past', 'present', 'future' and the relational predicate 'when.'
If this can be done then McTaggart's paradox cannot get off the
ground. To quote Ludlow
To illustrate, take a proposition like [] [3e) e is the dying of
Queen Anne] []
That proposition was future and is now past, but we can't over-
look the temporal anaphora. There is an implicit when-clause, so
that what we actually have is that the proposition was future (say,
when Queen Anne was born) and it is past (say, as I write these
words). There is not even the illusion of a contradiction if we
remember to include the temporal anaphora. (1999, 134)
The question I have concerning Ludlow's response to McTaggart's
argument is this: 'What is the ontological significance of the tenses
and the 'when' clause?
Ludlow notes that,
If this gambit is to work, 'when' cannot mean 'at the same time';
it must be taken as a kind of primitive, just as PAST, PRES, and
FUT morphemes are. That is, 'when' must be understood as
being more fundamental than the B-series conception of simul-
taneity. (1999, 112; my emphasis)
What does 'primitive' mean in this context? One would think that
in doing the metaphysics of time, to take something as primitive is
not only an epistemological notion, but an ontological notion as
well. When one is attempting to reduce temporal relations to some
ontologically more basic entity or entities, to take something as
primitive should imply ontological commitment to what is referred
to by that primitive term. A primitive is thus what is unanalysable
in terms of simpler constituents. However, for Ludlow primitives
do not name simple constituents of complex facts since he says,
'[M]y A-theory of tense will regard tenses as being predicates of
proposition like objects.' (1999, 112; my emphasis) And he claims
that predicates, including I would surmise, relational predicates
such as 'when' are non-referring expressions that do not denote
non-relational or relational properties construed either as Platonic
entities, or as extensions (i.e., sets of objects). Indeed, he maintains
that properties are 'very poor candidates for our ontology' (1999,
46) If, however, there are no A-properties referred to by the tenses,
and there are no temporal relations referred to by the language that

87
L. Nathan Oaklander

expresses them, then what exactly is Ludlow's ontology of time?

What is the ground, in the sense of the truth-makers, for the world
being temporal} Ludlow says that we can avoid taking 'times' as
being points in the sense of B-series metaphysics, by treating 'times
as sets of when-clauses' (1999, 128), that he claims are temporal
conjunctions. But he never explains what is that bit of reality that
makes a 'when' clause a temporal conjunction?7
Although Ludlow clearly seems to be partial to desert landscapes
he is, he says, 'still after a theory that delivers language-to-world
connections. The point here is that properties and extensions (sets
of objects) don't have to be part of that picture.' (1999, 46) Perhaps
not, but if properties and presumably relations are not to be included
in reality, then I fail to see how the rest of his very sophisticated
semantical theory can specify what there is in the world that is the
truth-maker of judgments asserting that two entities stand in a tem-
poral relation. Without such an account the question, Can an onto-
logical analysis that specifies the constituents of the complex fact
that it is now and the relevant event is earlier or later than now be
given that is consistent with presentism? has yet to be given an affir-
mative answer. And without an affirmative answer, our differential
attitudes toward earlier and later events remain a mystery on pre-
sentist metaphysics, and the existence of change remains without an
adequate foundation.8

References
Bergmann, Gustav 1964. 'Generality and Existence', in Logic and Reality.
Madison: University of Wisconsin Press, 64-84.
Bigelow, John 1996. 'Presentism and Properties', in James E. Tomberlin
(ed.), Philosophical Perspectives 10: Metaphysics, 1996. Cambridge, MA
and Oxford, UK: Blackwell: 35-52.
7
Ludlow believes that evidence for his view is gleaned from some psy-
chological studies that 'have noted that a particular sense of 'when'
emerges before the child has a notion of temporal order and simultaneity'.
([1999: 141) He mentions 'Cromer's (1968, 100) fascinating conclusion
that 'perhaps the ability to 'date' an event by a contemporaneous event is
more 'primitive' than the notion of serial ordering' (Ludlow 1999, 141),
but by appealing to a contemporaneous event, i.e., one occurring at the same
time, the study has not eliminated the B-relation of simultaneity, but has
presupposed it.
8
An earlier version of this paper was read at the Conference on Time,
Reality and Experience at the London School of Economics, in London,
on September 28, 2000.
88
 Presentism, Ontology and Temporal Experience

Broad, C. D. 1938. An Examination of McTaggart's Philosophy.

Cambridge: Cambridge University Press.
Craig, William Lane 1997. 'Is Presentness a Property?' American
Philosophical Quarterly 34, 27-40.
1998. 'McTaggart's Paradox and the Problem of Temporary
Intrinsics', Analysis 122-7.
1999a. 'Tensed Time and Our Differential Experience of the Past
and Future', Southern Journal of Philosophy 37, 4, 515—37.
1999b. 'Oaklander on McTaggart and Intrinsic Change', Analysis 59,
4, 319-20.
2000a. 'The Extent of the Present', International Studies in the
Philosophy of Sciences 14, 2, 165-85.
2000b. The Tensed Theory of Time—A Critical Examination.
Dordrecht: Kluwer Academic Press.
2000c. The Tenseless Theory of Time—A Critical Examination.
Dordrecht: Kluwer Academic Press.
2001. Time and the Metaphysics of Relativity. Dordrecht: Kluwer
Academic Press.
Cromer, R. F. 1968. The Development of Temporal Reference During the
Acquisition of Language. Ph.D thesis, Harvard University.
Grossmann, Reinhardt 1992. The Existence of the World—An Introduction
to Ontology. New York: Routledge.
Hinchliff, Mark 1996. 'The Puzzle of Change', in James E. Tomberlin
(ed.), Philosophical Perspectives 10: Metaphysics, 1996. Cambridge, MA
and Oxford, UK: Blackwell: 119-36.
Hochberg, Herbert 1969. 'Negation and Generality,' NOUS 3, 325^3.
Le Poidevin, Robin 1991. Change, Cause and Contradiction: A Defense of
the Tenseless Theory of Time. London: Macmillan.
(ed.) 1998a. Questions of Time and Tense. Oxford: Clarendon Press
1998b. 'Review of Michael Tooley's Time, Tense and Causation',
British Journal for the Philosophy of Science 49, 365-69.
1999a. 'Egocentric and Objective Time', Proceedings of the
Aristotelian Society. New Series 99, 19-36.
1999b. 'Can Beliefs Be Caused By Their Truth-Makers? Analysis 59,
3, 148-56.
2001. 'Reply To Smith and Tooley', in L. Nathan Oaklander (ed.),
The Importance of Time. Dordrecht: Kluwer Academic Publishers,
285-91.
Lewis, David 1986. On the Plurality of Worlds. Oxford: Blackwell.
Ludlow, Peter 1999. Semantics, Tense, and Time—An Essay in the
Metaphysics of Natural Language. Cambridge: MA: Massachusetts
Institute of Technology.
Mellor, D. H. 1998. Real Time II. Routledge: London.
Nerlich, Graham 1998. 'Time As Spacetime', in Robin Le Poidevin (ed.),
Questions of Time and Tense. Oxford: Clarendon Press, 119-34.
Oaklander, L. Nathan and Silvano Miracchi (1980), 'Russell, Negative
Facts, and Ontology', Philosophy of Science, 47, 434—55.
L. Nathan Oaklander

— 1984. Temporal Relations and Temporal Becoming: A Defense of a

Russellian Theory of Time. Lanham, MD: University Press of America.
— and Smith, Quentin (eds.) 1994. The New Theory of Time. New
Haven, Conn.:Yale University Press.
— 1996. 'McTaggart's Paradox and Smith's Tensed Theory of Time',
Synthese 107, 205-21.
— 1999a. 'Review of Michael Tooley's Time, Tense and Causation',
Mind 108, 407-13.
— 1999b. 'Craig on McTaggart's Paradox and the Problem of
Temporary Intrinsics', Analysis 59, 4, 314—18.
(ed.) 2002. The Importance of Time. Dordrecht: Kluwer Academic
Publishers.
Plantinga, Alvin 1983. 'Reason and Belief,' in Alvin Plantinga and
Nicholas Wolterstorff (eds), Faith and Philosophy. Notre Dame, IN:
University of Notre Dame Press, 39-63.
Prior, Arthur 1959. 'Thank Goodness That's Over', Philosophy 34, 12-17.
1967. Past, Present and Future. Clarendon, Oxford, UK.
1968. Papers on Time and Tense. Clarendon, Oxford UK.
1970. 'The Notion of the Present', Studium Generale 23, 245-48.
1996. 'Some Free Thinking About Time', in B. J. Copeland (ed.),
Logic and Reality, Essays on the Legacy of Arthur Prior. Oxford:
Clarendon Press.
Russell, Bertrand 1918. The Philosophy of Logical Atomism, in Robert
Charles Marsh (ed.) (1964), Logic and Knowledge: Essays 1901-1950.
London: George Allen & Unwin, 177—281.
Smith, Quentin 1986. 'The Infinite Regress of Temporal Attributions',
The Southern Journal of Philosophy 24, 383-96. Reprinted in L. Nathan
Oaklander and Quentin Smith (eds.), The New Theory of Time (1994).
1993. Language and Time. New York: Oxford University Press.
1999a. 'Review of Michael Tooley's Time, Tense and Causation', The
Philosophical Review 108, 123-27.
1999b. 'The 'Sentence-Type Version' of the Tenseless Theory of
Time', Synthese 119, 233-51.
2002a. 'Reference to the Past and Future', in Alexandar Jokic and
Quentin Smith (eds.), Time, Tense and Reference. Cambridge, MA:
MIT Press.
Smith, Quentin 2002b. 'Time and Degrees of Existence: A Theory of
"Degree Presentism",' in Craig Callender (ed.), Time, Reality and
Experience, Cambridge: Cambridge University Press, 119—36.
Tooley, Michael 1997. Time, Tense and Causation. Oxford: Clarendon
Press.
Zimmerman, Dean 1998. 'Temporary Intrinsics and Presentism', in Peter
van Inwagen and Dean Zimmerman (eds), Metaphysics, The Big
Questions. Oxford: Blackwell, 206-19.

90
A Presentist's Refutation of Mellor's
McTaggart
PHILIP PERCIVAL

For twenty years, D. H. Mellor has promoted an influential defence

of a view of time he first called the 'tenseless' view, but now
associates with what he calls the 'B-theory.'1 It is his defence of this
view, not the view itself, which is generally taken to be novel. It is
organized around a forcefully presented attack on rival views which
he claims to be a development of McTaggart's celebrated argument
that the 'A-series' is contradictory. I will call this attack 'Mellor's
McTaggart.' Although it has received much critical attention, it has
not been well understood. For one thing, it has changed over the
years in a way that is little appreciated. Whereas Mellor's (1981)
original version amounts to a dilemma each horn of which contains
a single strand of argument, later statements (1986, 1998) of the
first horn contain a second strand of argument unannounced. I
shall be concerned to disentangle these strands. I shall also show
them to have been largely anticipated by Gareth Evans (1979).
However, my main aim is not the clarification of Mellor's
McTaggart, but its refutation. I shall show that there is a rival to
Mellor's view of time against which the first horn of the dilemma
begs the question both as originally presented, and as supple-
mented. This rival is a 'Priorean' version of the 'presentist' doctrine
that only what is present exists. Although Prior himself gave
McTaggart's own argument short shrift, in refuting Mellor's devel-
opment of it I do not merely resurrect Prior's moves. Mellor's
McTaggart introduces specifically semantic considerations. It
focuses not as McTaggart did on presentness and futurity etc., but
on the truth-values of tokens of propositions in which presentness
and futurity etc. are (said to be) ascribed. Consequently, its
refutation requires an answer, from the perspective of a presentist
metaphysics, to a question which came to the fore only after Prior's
death. The question is this: How should semantic theory be
developed in the light of the need for a theory of linguistic under-
standing? Though Evans (1979) flirted with the issue of how this
question should be answered from a presentist perspective, the

1
See, respectively, Mellor (1981) and Mellor (1998).
91
Philip Percival

answer he articulates is wrong. I shall do no more than sketch the

correct answer. That is all a refutation of Mellor's McTaggart
requires.
Before Mellor's McTaggart can be refuted (section III), it must
first be clarified. This is a matter of clarifying the conclusion
Mellor wishes to establish (section I), and the argument by which
he tries to establish it (section II). Clarifying the conclusion to
Mellor's McTaggart is in some ways the hardest task. It is a messy
and laborious business, and I bid the reader to be patient.

I 'Tensed' vs. 'Tenseless' views of time

Li A (philosophical) defect in Mellor's terminology—Mellor (1981: 4)

opposes a 'tenseless' view of time to a 'tensed' view, and then claims
that his semantic development of an argument due to McTaggart
reveals the tensed view to be inconsistent. His later dissatisfaction
with this terminology stems from the fact that whereas the terms
'tensed' and 'tenseless' apply primarily to language, the debate he
has in mind concerns the nature of (temporal) reality. Still, although
his earlier terminology is potentially misleading in this respect,
Mellor's (1998: xi) current employment of the alternative terms 'B-
theory' and 'A-theory' redeems matters only at the cost of an
unpalatable blandness. I will revert to Mellor's original talk of
'tensed' and 'tenseless' views of time.
According to the tensed view of time, 'tense' is real. The tense-
less view denies this: it holds tense unreal. But what is it the reality
of which is thereby disputed? Primarily, 'tense' is a feature of
language: verbs can take a variety of grammatical forms, or 'tenses.'
With respect to this primary meaning there can be no question of a
metaphysical debate over 'tense': tense in this sense is a real feature
of language. However, if grammatical tense is not the issue, perhaps
it is a guide to what is. In particular, one might wonder whether
grammatically tensed sentences, or uses of such sentences, success-
fully represent the world as having a certain feature, to be called
'tense.' The debate between the two views of time might then be
taken to dispute not the reality of grammatical tense, but its
function, and what the successful exercise of that function involves.
Perhaps the tensed view should be characterized as holding that
grammatical tense successfully serves a representational function—
representing a feature, to be called 'tense,' which (some part of)
reality possesses—while the tenseless view should be characterized
as denying this. However, though this characterization would be
92
 A Presentist's Refutation of Mellor's McTaggart

closer to the mark, it would be fundamentally wrong. Firstly, even

if the debate between the tensed and tenseless views may be
characterized as concerning the representational function of certain
sentence-types, these sentence-types are not singled out by their
employment of grammatical tense: containing a grammatically
tensed verb is neither necessary nor sufficient for being of the type
in question. Secondly, the debate over the reality of tense should not
in any case be characterized as turning on whether the world has
some feature, 'tense,' which the tensed view takes to be represented
by (uses of) the disputed sentence-types. To understand the debate
between tensed and tenseless views of time, one needs to under-
stand why these two misconceptions are such.
Let's begin with the first misconception. According to Mellor
(1981, 1986), the disputed sentence-types are distinguished not by
their employment of grammatically tensed verbs, but by the fact of
their being 'tensed.' But what is it for a sentence-type to be 'tensed'?
Having said that
TENSES ... [are] ... temporal positions in McTaggart's A series,
'that series of positions which runs from the far past through the
near past to the present, and then from the present through the
near future to the far future,'
Mellor (1986: 167) answers this question by saying that
[A] thought, statement or sentence ... [is] ... TENSED if it
explicitly or implicitly ascribes a TENSE to something, its
TENSE being the TENSE is ascribes.2
However, this characterization of the sentences of the dispute class
—the 'tensed' ones—is disastrous. Firstly, since the tenseless view
of time holds that in a non-grammatical sense, 'tense' is unreal, it is
wildly obfuscatory to begin an argument to this conclusion by
insisting that there are entities which, in a non-grammatical sense,
are 'tensed.' Secondly, the characterization of 'tensed' representa-
tions as those which 'ascribe' a 'tense' to something conflicts with
Mellor's own view of time. His view is not an 'error-theory.' The
view that 'tense' is unreal is not the view that all token utterances of
sentence-types like e.g. 'The great day is finally present' are false:
on the contrary, utterances of this sentence on the (great) day in
question are held to be true. Yet if the sentence ascribes a tense to
2
Cf. Mellor (1981: 4), and Le Poidevin and Mellor (1987). Mellor
(1986) only capitalizes 'tense' to emphasize its distinctness on his usage
from grammatical (verbal) tense. In discussing his views I drop this
practice.
93
Philip Percival

something, when in reality nothing has any tense, no utterance of it

could be true: ascribing something which nothing possesses results
in falsehood.3 Thirdly, allowing a representation to be 'tensed' even
if it ascribes a tense to something only 'implicitly,' and then pre-
supposing, as Mellor does, that there are representations, i.e. the
'tenseless' ones, which are not tensed, begs the question. In the case
of sentences (rather than representations of other kinds), the
contrast he intends is between sentences like 'At last the exam is in
the past' and 'Bush ate fish yesterday' on the one hand, and
sentences like 'World War II exists in the past, or the present, or the
future' and 'Bush eats dates at noon on Christmas Day, 2004' on the
other. But while sentences of the latter kind do not specify the tense
of anything explicitly, it is a moot point whether they do so
implicitly. Suppose, as some think, that future entities are
unnameable. In that case, wouldn't any sentence containing 'World
War II' 'implicitly' say that World War II is either past or present?
Or again, suppose, as some think, that future contingents lack truth
values. In that case, wouldn't even an 'eternal' sentence like e.g.
'Bush reads at t' 'implicitly' say that t is past or present?
Let's leave the issues raised by the unsatisfactoriness of Mellor's
characterization of 'tensed' representations for the moment and
turn to the second misconception. I claim that the debate between
the tensed and tenseless views should not be construed as a debate
over whether there is a feature of the world—'tense'—which
commonplace talk of temporal matters often represents (parts of)
the world as having. This claim might be thought presumptuous,
since it flies in the face of Mellor's explicit definitions to the
contrary. Having characterized the 'tense' of something other than
a representation as its A-series position, in Real Time Mellor says
(p. 4) that the tensed view holds, whereas the tenseless view denies,
that
Distinctions and transitions of tense ... reflect nonrelational
differences between past, present and future things (events, facts,
etc.) ... Futurity, temporal presence and pastness ... [are] ... real
nonrelational properties which everything in time successively
possesses, changing objectively as it exchanges each of these
properties for the next ... [properties which] ... are ... much like
(e.g.) temperatures.
3
This objection is not undermined by Mellor's (1998) new terminology.
It is equally applicable to a strategy which begins an argument for a non
error-theoretic view that there are no 'A-facts' by characterizing sentences
or propositions etc. belonging to the class of 'A-sentences or A-
propositions' as those which explicitly or implicitly state A-facts.
94
 A Presentist's Refutation of Mellor's McTaggart

In Real Time II, the debate between the 'A-theory' and the 'B-
theory' is characterized similarly (p. 2):
The question is, what makes a statement like 'e is past' true when
it is true, namely at any time later than e ? There are two answers
to this question. One is that at any such time e has the property
of being past. This is what I call the 'A-theory' view. My own 'B-
theory' view is that what makes 'e is past' true at any time t is the
fact that e is earlier than t.
To be sure, Mellor is setting his own agenda, and introducing novel
technical terms to describe it. Nevertheless, this isn't to say he can
define those terms as he pleases. In defining them he is trying to
capture an opposition between the view of time he subsequently
defends, and a view of time which he takes to be refuted by his
development of McTaggart's argument against the reality of the A-
series. His initial characterization is therefore subject to a retro-
spectively imposed norm: it should characterize these views. But it
doesn't, and to this extent his official characterization of the debate
at the heart of Real Time (and its successor) is mistaken. There are
two reasons why this is so.
Firstly, as Mellor knows all too well, no one opposes his view of
time more strongly than Arthur Prior. Yet Prior (1967: 18) holds
that '"Is present", "is past", etc. are only quasi-predicates,' which
do not designate properties ('like temperatures'), and which are to
be analysed by means of sentential operators like 'it was the case
that.' Prima facie, therefore, on Mellor's official characterization,
the tenseless view he is concerned to promote is embraced by his
most formidable opponent! Mellor (1981: 95; cf. 1998: 75) tries to
avoid this anomaly by dismissing the significance of Prior's
employment of temporal (sentential) operators on the grounds that
it is 'tantamount to regarding P[astness], N[owness] and F[uturity]
as properties, not of events, but of tensed facts.' If this were so, his
characterization of the tensed and tenseless views would indeed
classify Prior's metaphysics correctly. Despite holding that e.g. 'is
past' is a 'quasi-predicate,' which does not designate a property of
events, Prior's employment of temporal (sentential) operators
would commit him to the view that e.g. pastness is a non-relational
property (of some facts), and, hence, to the reality of tense and the
tensed view of time. However, Mellor's diagnosis of Prior's
employment of temporal operators is too controversial to bear the
burden being placed upon it. In part, this is because an ontology of
facts is itself controversial. But even if this ontology is admitted,
Prior's doctrine of temporal operators has good reason to resist

95
Philip Percival

Mellor's recommendation. On Mellor's diagnosis, the metaphysical

import of Prior's employment of temporal operators has e.g. the
joint truth of 'Bush is reading' and 'It was the case that (Clinton is
reading)' requiring the existence of two facts, the second being
different from the first in two respects: it involves Clinton rather
than Bush, and it possesses not the property of being present, but
the property of being past. But Prior would resist this attribution as
strongly as I (and many others) would resist the idea that the joint
truth of 'Bush is reading at noon on Christmas Day, 2002' and 'It
might have been the case that (Bush is walking on the moon at noon
on Christmas Day, 2002)' requires the existence of two facts, the
second differing from the first in two respects: it involves walking
where the other involves reading, and it possesses the property of
being possible instead of the property of being actual. One might
object that in the absence of a positive account of what a temporal
operator like 'It was the case that' does if it does not designate a
property of facts, an attempt to resist Mellor's attribution fails.
However, I think it wrong to insist on forcing sentential operators,
temporal or otherwise, into the mould of the classical ontology of
entities and their properties and relations. Nor is Mellor any longer
in a position to insist on it. In replacing Real Time's view that e.g.
'Bush is reading at t' has the form 'Reading (Bush, t)' by Real Time
IPs view that it has the form 'at t (Bush, reading),' he now invokes
sentential operators which cannot be forced into this mould.4 (On
his latest view, 'at t' designates not a property (of a fact) but a
'temporal location' (of a fact).)
Mellor intends a debate on which the 'tenseless' view of time is
essentially opposed to Prior's view. Prior's denial that e.g. 'pastness'
is a property ('like temperature') affords one reason for seeking an
understanding of the claim that 'tense is real' on which tense can be
held real even though it is not a non-relational property, and, a
fortiori, not a property determinations of which are designated by
e.g. 'is past' or 'it was the case that.' A second reason for seeking
such an understanding is provided by Real Time H's rewriting of
Real Time's discussion and development of McTaggart's argument
to include a discussion of Tooley's (1997) view of time.5 Because
Tooley's view does not advert to non-relational properties of
presentness etc., Mellor (1998: 81) classifies it, using his new termi-
nology, as a 'B-theory.' Yet Tooley's view is also held to be refuted
by the semantic development of McTaggart's argument against the
4
5
See respectively chapters 7 and 8 of Mellor (1981) and Mellor (1998).
See chapters 7 and 6 respectively of these works. The additional
material in question is to be found in Mellor (1998: 81-3).
96
 A Presentist's Refutation of Mellor's McTaggart

reality of the A-series. Once more, then, Mellor's central thesis, as

defended by the McTaggartian argument which is the lynchpin of
both Real Time and its successor, is seen to be stronger than the
thesis that there are no non-relational properties of pastness,
presentness and futurity by which the 'tenseless' view of time is
officially characterized.
Mellor's characterization of the thesis that tense is real, and
hence of the tensed view of time, as the thesis that there are entities
which possess temporally variable non-relational properties of past-
ness, presentness and futurity, captures an important object of
debate. But his dispute with Prior and Tooley, and his employment
of Mellor's McTaggart against them, reveals this debate to be less
deep than the one at the real heart of Real Time and its successor.
Although one might retain Mellor's characterization of the thesis
that 'tense' is unreal, and hence of the debate between the 'tensed'
and 'tenseless' views of time, and introduce new terminology to
capture the deeper one, so doing would miss something philosoph-
ically. In effect, in Real Time the phrase 'the tenseless view' is used
in two ways that are assumed to coincide. As characterized officially,
it signifies the thesis that nothing has non-relational properties of
pastness, presentness and futurity. But it is also employed to
signify a view which this thesis is implicitly assumed to entail,
notwithstanding the fact that this latter view explicitly conflicts
with the respective views of Prior and Tooley. In recognizing this
assumption, and including additional material in which a view like
Tooley's is critically discussed for the first time, Real Time II is
obliged to choose, in effect, between Real Time's two uses of 'the
tenseless view' As we have seen, when introducing the term 'B-
theory' as an alternative to 'the tenseless view,' Mellor chooses to
retain the official characterization. But this choice results in the
absurdity of there being no term in Real Time II for the (stronger)
view of time which is carried over from its predecessor, and which
is now explicitly defended by the central (McTaggartian) argument
of the two books.
A better alternative corrects this anomaly. The phrase 'the tense-
less view of time' should be employed to signify a view of time
which (i) Real Time took it to signify (albeit not officially), which (ii)
is defended in Real Time and its successor, and which (iii) Real Time
implicitly assumes, and Real Time II explicitly argues, to be proven
correct by Mellor's McTaggart. This view of time holds more than
the weaker thesis that nothing in reality has non-relational
properties of pastness, presentness or futurity. Real Time II's term
'B-theory' can be retained, as defined by Mellor, for this weaker
97
Philip Percival

thesis. The 'tenseless' view of time then amounts to 'B-theory' plus

something extra.
The next question is: what is the additional feature which turns a
mere B-theory into the full-blown tenseless view of time?

I.it The tenseless view—Cursory reflection reveals the crucial

additional feature in virtue of which Mellor (1981) takes his
'tenseless' view to be stronger than his own official characterization.
Like the explicit conclusion of the semantic development of
McTaggart's argument against the reality of the A-series in Real
Time II, it is characterized most generally not as the rejection of
non-relational properties of presentness etc., but as the rejection of
a view of (a kind of) temporal change, and hence of temporal reality,
which is entailed by, but does not require, the existence of such
properties. In terms of an ontology of (truthmaking) facts, the
rejected view is easily stated. It is simply the view that the facts
change over time.6 Since an ontology of facts would suggest, as in
the Tractatus, that reality comprises the facts, this view quickly
becomes the view that reality itself (so to speak, conceived as a
whole) changes over time.
Following the terminological recommendation of the previous
section, temporally changing facts can be called 'tensed' facts;
unchanging ones, 'tenseless' facts. The terms 'A-fact' and 'B-fact'
can then be retained as defined in Real Time II: 'A-facts' are facts
which involve, or which possess, non-relational properties of past-
ness, presentness, or futurity. 'B-facts' neither involve, nor possess,
such properties. Clearly, if there are A-facts, then there are tensed
facts. But the status of the converse entailment is unclear. Prima
facie, temporal change in the facts needn't involve 'A-facts.' If it
does involve them, it might perhaps be a matter of temporal
variation in the properties of eternally existing A-facts: after all, as
we have seen, Mellor himself attributes to Prior the idea that
futurity, presentness, and pastness are properties which facts
successively possess. More plausibly, however, it might be a matter
of facts coming to exist or ceasing to exist (i.e. the presumption
being that existence is not a property). Again, one variant of this
alternative holds that the facts which do this include such A-facts as
6
Cf. Mellor's (1981: 103) claim that '[Because] tenses cannot change ...
The world canfnot] ... grow by the accretion of facts as they become
present.' See too Mellor's (1998: 81) claim that 'McTaggart's proof
disposes of more than A-facts [i.e. facts involving non-relational properties
of pastness, presentness or futurity]. It disposes also of the idea ... that
even B-facts need not exist at all times.'
98
 A Presentist's Refutation of Mellor's McTaggart

an event's having a non-relational property of presentness: such a

fact would not have existed until the event became present, and
would cease to exist when the event becomes past. But other
variants hold that even B-facts come into, or go out of, existence, or
both, with the passage of time. On Tooley's view for example, the
passage of time amounts to the successive expansion of reality as
new B-facts of this kind come into, only then to remain forever in,
existence. By contrast, someone sympathetic to Dummett's (1969)
anti-realism about the past might have facts ceasing to be with the
passage of time (as they cease to be knowable), but never coming to
be (since mere futurity cannot undermine the following decision
procedure: wait around). 7 Finally, a Priorean presentist might have
a fact—e.g. Bush is sitting down—successively coming and ceasing
to be as Bush goes about his daily business.8
By contrast, on the view of time which Mellor defends, and which
he takes his development of McTaggart's argument against the
reality of the A-series to establish, the facts don't, and hence reality
(as a whole) doesn't, change over time in any of these ways. Mellor's
official characterization notwithstanding, this is what Real Time
eventually uses the phrase 'the tenseless view' to signify. Since Real
Time II both continues to uphold this view of time, explicitly, and
to take Mellor's McTaggart to establish it, it is a pity that the
official characterization is retained in the terminological shift to 'B-
theory.' The consequence that neither Real Time nor its successor
contain a term by which their view of time is signified stems from
Real Time's presumption that if there are no non-relational
properties of futurity, presentness or pastness, then the facts don't
change over time. Judging by the rewriting of Mellor's McTaggart
in Real Time II, Tooley opened Mellor's eyes in this respect. But a
more sensitive reading of Prior might have opened them earlier.
It might seem that we are in a position to define the 'tenseless'
view of time: it is B-theory's rejection of non-relational properties
of pastness, presentness, or futurity, plus the view that B-facts are
7
See too Dummett (1981: 382-400). The relation between Dummett's
debate regarding realism about the past and Mellor's debate regarding the
tenseless view is considered in Yuval Dolev (2000).
8
As we have seen, Prior resists properties of pastness, presentness or
futurity. On this conception, 'Bush is sitting down' would be a B-fact
which lacks times as constituents. Mellor (1998) himself advocates B-facts
of this kind. But his view is very different from Prior's. Mellor takes these
B-facts to be unchanging, and gives them unchanging 'temporal locations'
'at' dates. Given an ontology of facts, Prior would take e.g. 'Bush is sitting
down' to be a changing fact which has no temporal location. (For Prior,
'dates' are logical constructions out of propositions.)
99
Philip Percival

unchanging over time. However, there is reason to be cautious at

this juncture, since an ontology of facts is controversial: there are
those who deny the existence of facts.9 We should at least try to find
an alternative characterization of the 'tenseless' view from which
Mellor's ontological presupposition is absent.
The obvious alternative, which Mellor's own practice encourages,
is to couch the debate in terms of truth conditions.10 For there is
hope that talk of truth conditions does not presuppose novel enti-
ties. Truth conditions are said to be possessed by sentences, or by
utterances of sentences, and to 'obtain.' But for 'snow is white' to
have the 'property' of having the truth condition it does might
amount to nothing more than that according to the 'canonical'
theorem for 'snow is white' in the theory of truth for English which
does best by certain criteria, 'snow is white' is true iff snow is white.
Likewise, for this truth condition to 'obtain' might amount to
nothing more than that snow is white. So we can adopt the follow-
ing proposal. The dispute between A-theory and B-theory turns on
whether the truth conditions of tokens of propositions like 'The
great day is present' or 'Bush is sitting down' involve non-relational
properties of presentness, pastness or futurity: A-theory says they
do, while B-theory denies this. The 'tenseless' view of time then
adds to B-theory the claim that the truth conditions of token
propositions are 'tenseless' in the sense that the obtaining of such
tokens' truth conditions is not a matter which varies over time. The
'tensed' view of time denies this: it holds that some token
propositions have truth conditions that are 'tensed' in so far as
whether their truth conditions obtain is a matter which varies over time.
In what follows, I focus on this latter formulation. In its terms,
Mellor takes his semantic development of McTaggart's argument
against the reality of the A-series to show that no token proposition
can have tensed truth conditions.

II Mellor's McTaggart
Il.i The original version—Mellor's McTaggart (1981) is best articu-
lated as a dilemma. Mellor himself develops the dilemma with
9
Cf. Lowe (1987b: 539) who declines to enter the dispute over whether
there are tensed 'facts' on the grounds that he 'can find little use for this
notion.'
10
This is the focus of the discussion in e.g. Lowe (1987a, 1987b, 1992);
Mellor (1981, 1986); Paul (1997); Priest (1986); and Smith (1994).
(However, Mellor himself doesn't see truth conditions as an escape route
from an ontology of facts: for him, the truth condition of something is that
some fact obtain.)
100
 A Presentist's Refutation of Mellor's McTaggart

respect to a sentence-type 'e is past' (where e is some event), but so

as to emphasize that the tensed view which is really being attacked
is not committed to the supposition that there are non-relational
properties of pastness, presentness, and futurity, I prefer to use an
atomic sentence-type containing a present tense verb, such as 'Bush
is sitting down.' Nothing of substance, except greater generality,
turns on this choice. Many will agree that if any sentence-type is
such that tokens of it have truth conditions the obtaining of which
is a matter which varies over time, this one does. Those who deny
this can substitute some other sentence-type they deem a better
candidate for the distinction of having tensed truth conditions. The
argumentation of Mellor's dilemma is structural, and they will have
no difficulty in applying it to the sentence-type of their choice.
For any sentence-type S, let S be 'non-indexical with respect to
time' if the truth conditions of a token utterance of S do not depend
on that token's temporal location, and let S be 'indexical with
respect to time' otherwise. 'Bush is sitting down' is either indexical
with respect to time or it is not. Mellor's McTaggart (1981) as
originally developed first argues that the tensed view of time cannot
accommodate the thesis that this sentence-type is non-indexical
with respect to time. It then argues that the tensed view of time
cannot accommodate the thesis that this sentence-type is indexical
with respect to time. More particularly, Mellor (1981) propounds
the following dilemma:

First Horn—Suppose 'Bush is sitting down' is non-indexical with

respect to time. Since it contains no non-temporal indexicals, this is
to say that all of its token utterances have the same truth conditions.
In that case, all of its tokens have the same truth-value irrespective
of their temporal locations. But this is impossible. Since what
inclines the tensed view to say that utterances of this sentence-type
have tensed truth conditions in the first place is recognition of the
fact that this sentence-type is temporally sensitive, and, hence that
the truth-value of a token utterance of it depends in part on the
token's temporal location, it would result in some token utterances
ending up both true and false. It is therefore incoherent to suppose
that 'Bush is sitting down' is non-indexical with respect to time."
" Compare the statement in Mellor (1981: 100-1) that 'any attempt to
state in a tensed meta-language the one tensed fact that makes all ... true
tokens [of "e is past"] true is bound to fail. The alleged fact would by
definition have to make all tokens of the type true, regardless of their A-
series position, whereas in fact some are always true and others always
false.'
101
Philip Percival

Second Horn—Suppose 'Bush is sitting down' is indexical with

respect to time. In that case, the truth conditions of token
utterances of it are given by some such clauses as12

Date—A token u of the sentence-type 'Bush is sitting down' uttered

at t [is] true if and only if Bush [is] sitting down at t

or

Token-Reflexive—A token u of the sentence-type 'Bush is sitting

down' [is] true if and only if u [is] uttered simultaneously with
Bush's-sitting-down.

However, the truth conditions such clauses specify are tenseless:

Whether such truth conditions obtain is not a matter which varies
over time. For each time t, whether or not Bush is sitting down at t
is not a matter which changes over time; for each utterance u,
whether u is simultaneous with Bush's-sitting-down is not a
temporally variable matter. The thesis that the sentence-type 'Bush
is sitting down' is indexical with respect to time is therefore
incompatible with the supposition that token utterances of it have
tensed truth conditions.13

II.ii Later versions—The semantic development of McTaggart's

argument against the reality of the A-series in chapter 6 of Real
Time has recently appeared, rewritten, as chapter 7 of Real Time II.
In a preface, Mellor acknowledges many revisions, supplements,
and changes of opinion. In particular, he relates that the rewriting
of Real Time's chapter 6 contains an additional section which
attempts to refute what is said to be 'Michael Tooley's (1997) ...
theory ... [which] admits only B-facts while denying ... that they
exist at all times.' In the terminology I have recommended, this
additional section supplements the second horn of the original
dilemma: it amounts to a further argument to the effect that if e.g.
12
In placing certain occurrences of 'is' in the clauses which follow
within brackets, I follow Lowe (1998). My reasons for so doing will
become clear in a moment. I am enriching Mellor's (1981) second horn a
bit, since that argument is committed to the Token-Reflective account and
does not recognize the alternative, which I have called 'Date,' by which it
is replaced in Real Time II. I do so because the switch from 'Token-
Reflexive' to 'Date' does not affect the original second horn's form of
argument, and is therefore relatively superficial.
13
Cf. Mellor (1981: 101).
102
 A Presentist's Refutation of Mellor's McTaggart

the sentence-type 'Bush is sitting down' is indexical with respect to

time, the truth conditions of its tokens, or the facts in virtue of
which true tokens of it are true, are indeed tenseless. Since I am
concerned to undermine the first horn of Mellor's dilemma, I won't
go into this supplement to the second one.'4
However, although Mellor does not acknowledge the fact, Real
Time II also contains additional material which supplements the first
horn. The argument in this material emerged in Mellor's (1986)
interchange with Graham Priest. While Priest (1986) mistakenly
reads it into Real Time's presentation of Mellor's McTaggart,
Mellor's (1986: 170-1) reply happily takes it on board and endorses
it. Curiously, Mellor (1988: 80-1) returns to form and omits it from
his statement of the first horn of his dilemma. But with Mellor's
(1998: 78) publication of Real Time II it is thoroughly assimilated:
[C]onsider two tokens, a and b, of 'e is past', one earlier than e and
one later. Suppose for example that e is Jim's race [at 4.30] on 2
June ... Then if a and b are both made true by the A-fact that e
is past, they must both be true when this is a fact and false when
it is not. So at 4p.m.. when e is still future, a and b must both be
false: and at 5p.m.. when it is past, they must both be true ... Yet
[these truth values] are obviously wrong. To say before Jim's race
that it is past is to produce a token of 'e is past' that is and always
will be false. Similarly, to say after his race that it is past is to pro-
duce a token that is and always was true ... Once we distinguish
propositions from their tokens, it is obvious that tokens of an A-
proposition, unlike the proposition itself, do not change their
truth values over time (my underlining).
As presented originally, the first horn of Mellor's McTaggart
claims that supposing a sentence like e.g. 'Bush is sitting down' to
be non-indexical with respect to time has the absurd consequence
that all tokens of this sentence-type, whatever their temporal
location, have the same truth value. I will call this the argument
'within' time. By contrast, this later presentation adds the further
claim that this supposition has the absurd consequence that each
14
Lowe (1987a,b; 1992, 1998) attacks the second horn of Mellor's
McTaggart, arguing that it is wrong to presume, as Mellor does, that the
truth-conditions of token utterances which are specified by Date or
Token-Reflexive are tenseless: on the contrary, he suggests, Date and
Token-Reflexive specify tensed truth-conditions provided the occurrences
of '[is]' they contain are read not as tenseless verbs of the same kind as the
one which occurs in e.g. 'two plus five is seven,' but as a disjunction 'is,
was, or will be' of tensed verbs (Lowe 1998: 44—5).
103
Philip Percival

token is such that its truth value switches between truth and falsity
over time. That is:
The supplement to the first horn—Suppose 'Bush is sitting down' is
non-indexical with respect to time. In that case, since it contains no
non-temporal indexicals, all of its token utterances have the same
truth conditions. It follows that the truth condition of a token
utterance of this sentence-type must be (something like) that Bush
is sitting down. So it must be a truth condition the obtaining of
which varies over time in the following way: with the progression
of time, it obtains, it does not obtain, it obtains ... If the obtaining
of a token utterance's truth condition varies over time in this way,
so too must its truth value switch between true and false. But this
is absurd. It contradicts the basic semantic fact that the truth value
of a token utterance does not vary over time in this way.
I will call this the argument 'over' time.15

II.Hi An untenable response—Priest (1986) accuses the first horn of

Mellor's McTaggart of begging the question. To assume that the
truth-value of a token utterance of a sentence-type like 'Bush is
sitting down' depends on the token's temporal location is, he
suggests, to take the tenseless view for granted. However, Mellor's
(1986) immediate reply that this just won't do is surely correct. It
1S
This development in the first horn of Mellor's dilemma has not
received the recognition it deserves, and the relation between the two
strands of argument is not always properly appreciated. As we have seen,
Priest conflates the two strands of argument, and in a way Mellor does too.
Robin Le Poidevin is also a case in point. Le Poidevin (1991; chs. 2—3)
holds that Mellor's (1981) argument against the tensed view of time
illuminates McTaggart's Paradox (1927), that only (Prior's) doctrine of
'presentism' (or 'temporal solipsism' as Le Poidevin calls it) can resist this
argument, and that presentism is nevertheless refuted by an argument in
Evans (1979). So, since the latter argument is precisely the argument 'over'
time, by which Mellor (1986, 1998) supplements Real Time's original
presentation of the first horn, in my terms Le Poidevin (1991) claims that
the argument 'within' time illuminates McTaggart, that it can be resisted
only by presentism, and that presentism is refuted by the supplementary
argument 'over' time. Yet despite reiterating his earlier claim that presen-
tism provides 'the only way for the tensed theorist to escape McTaggart's
paradox' when he returns to the topic, Le Poidevin (1998) imports the
supplementary argument into Mellor's McTaggart. (Since he implies that
he has not changed his mind either as to the force of the supplementary
argument, or as to the degree to which Mellor's argument illuminates
McTaggart's, this is at best disconcerting and at worst contradictory.)
104
 A Presentist's Refutation of Mellor's McTaggart

ought to be clear as day that the truth-value of a token utterance of

e.g. 'Bush is President of the U.S.' depends on the token's temporal
location. Everyone should agree that, pragmatic features aside, all
goes well when Jones speaks the words 'Bush is President of the
U.S.' during Bush's Presidency. But it is equally uncontroversial
that all did not go well when Jones spoke those words in November
1997, and to insist as Priest does that no harm is done if the two
utterances are given the same truth value is damningly obfuscatory.
As Mellor (1986) observes, the link between the truth of an utter-
ance and its correctness should be preserved. Semantically, the two
utterances of 'Bush is President of the U.S.' are not on a par.
Similarly, it just won't do to suggest, as Priest (1986) does, that
nothing is untoward if a semantic theory has the consequence that
the truth value of some utterance of e.g. 'Bush is sitting down'
switches between truth and falsity with the passage of time. The
link between the truth of an utterance of some sentence-type, and
the correct use of that sentence-type, is obvious, and one cannot just
accept, as Priest's proposal would have us do, that whether or not a
specific token utterance involves a correct use of the sentence-type
employed switches over time. As Evans (1979) points out, prima
facie, if the normative status of utterances one might make is tem-
porally variable in this way, one would be at a loss as to what to say.

II.iv An alternative strategy—Pace Priest, the difficulty with the

first horn of Mellor's dilemma is not that a question is begged when
certain consequences are deemed to be incoherent: these
consequences are indeed absurd. Rather, the difficulty is that, pace
Mellor, these consequences do not follow. Specifically, the mere
supposition that e.g. 'Bush is sitting down' is non-indexical with
respect to times need not have either of the (absurd) consequences
respectively claimed by the arguments 'within' and 'over' time: it
need have neither the consequence that all tokens of this sentence-
type have the same truth value, nor the consequence that the truth-
value of a token of this type changes from true to false, and vice-
versa, as Bush goes about his daily business. As I am about to
explain, the view of time on which this supposition lacks these
consequences is a 'Priorean' presentism.

III. Mellor's McTaggart refuted

Ill.i Prior's presentism—According to Arthur Prior, 'the present

simply is the real considered in relation to two species of unreality,
105
Philip Percival

namely the past and the future." 6 This is a version of 'presentism,'

the thesis that only what is present exists.17 Whatever other versions
of presentism might say, this version is not the thesis that there is a
non-relational property, 'presentness,' which all and only existing
things possess. Rather, on my understanding (at least), it couples
the thesis that everything that exists is picked out by the phrase
'everything that presently exists' with the thesis that pastness and
futurity are significant operations on existence, in that (i) what
exists might not have existed before and might not exist
subsequently; and, more contentiously, (ii) what exists might
include neither what did exist nor what will exist.
The key to understanding Prior's presentism is provided by his
sympathy for the scholastic doctrine that
'Socrates is sitting' is a complete proposition, enuntiabile, which
is sometimes true, sometimes false; not an incomplete expression
requiring a further phrase like 'at time t' to make it into an
assertion.18
For it is but the shortest of steps from this doctrine to the thesis that
temporality is an operation on propositions expressed by sentential-
operators like 'it was the case that' etc., and hence to the discipline
of tense-logic.
In effect, the scholastic doctrine of propositions and the doctrine
of temporal operators together announce that Prior declines to
think of temporality in the way that David Lewis thinks of
modality. David Lewis views a true modal sentence like '0(There
are talking donkeys)' as being true in virtue of the same sort of
thing—namely, talking donkeys (albeit non-actual ones)—as the
sort of thing—namely, (actual) donkeys—that a true atomic
sentence like 'There are donkeys' is true in virtue of. But most of us
do not think like this. We think of the modal sentence as being true
in virtue of something altogether different—the existence of
abstract talking donkeys, or of certain things doing duty for talking
donkeys, or of certain things representing talking donkeys, if we are
anti-modalists and either Ersatzers or fictionalists about possible
worlds, or, if we are modalists, the existence of some irreducible fact
(speaking loosely) of the form: 0(There are talking donkeys). Pace
Lewis, for us the actual simply is the real considered in relation to
one species of unreality, namely the (merely) possible.
Prior thinks of temporality, and hence of that in virtue of which
16
Prior (1970: 245).
17
Cf. Bigelow(1996: 35).
18
Prior (1967: 15-16). Cf. Geach (1949).
106
 A Presentist's Refutation of Mellor's McTaggart

temporal propositions like 'it will be the case that (there are
conscious computers)' and 'it was the case that (there are
dinosaurs)' are true, in the same way most of us think of modality.
These propositions are not held to be true, if true, in virtue of the
sort of thing that the atomic sentences 'there are computers' and
'there are reptiles' are true in virtue of. Rather, they are held to be
true in virtue of the existence of something doing duty for, or
representing, conscious computers and dinosaurs. Or better still,
they are held to be true in virtue of such brute, irreducible facts as
that: it will be the case that (there are conscious computers) and it
was the case that (there are dinosaurs).
Accordingly, Prior's presentism is the temporal analogue of a
modalism about possibility which is by no means unattractive. In
holding that pastness and futurity are operations on propositions, it
can deny that presentness, pastness and futurity are properties. It
might also endorse e.g. Lowe's (1987b: 539) suggestion that an
ontology of facts is obscure and by no means essential to the tensed
view.
Prior (1967: ch. 1) himself had little time for McTaggart's
paradox. He saw McTaggart's claim that if time exists, every event
is past, present, and future as a crass falsehood encouraged in part
by McTaggart's failure to recognize that complex tenses are
expressed by sentential operators, not by predicates. However, Prior
himself had no occasion to respond to the charge that it is
incoherent to suppose that token utterances of temporally sensitive
sentence-types" have tensed truth conditions, and while more
recent authors who favour the tensed view have shown some
sympathy for Prior's metaphysics, I know of no attempt to develop
a Priorean account of the truth conditions of token utterances
which expressly and self-consciously rebuts Mellor's semantic
development of McTaggart's argument.20
19
I.e. sentence-types the correct use of which is sensitive to time of
utterance.
20
I came close to so doing in an earlier paper (Percival (1989)). In
trying to build on what I took to be natural thoughts about the modal
analogue of Mellor's McTaggart, I suggested that Prior might try to resist
Mellor's dilemma by distinguishing evaluations of token utterances for
'truth' from evaluations of them for 'truth-at-a-time.' What follows
develops and corrects those inchoate thoughts. See too Percival (1990). For
further exploration of the modal analogue of the debate between the
tenseless and tensed views of time, and for discussion of the possibility of
arguments which distinguish the temporal and modal cases, see Percival
(1991, 1992, 1994).
107
Philip Percival

III.it Evans's anticipation of Mellor—How then might one who

thinks that
'Socrates is sitting' expresses a complete proposition, enuntiabile,
which is sometimes true, sometimes false; not an incomplete
expression requiring a further phrase like 'at time t' to make it
into an assertion
approach the task of giving truth conditions for token utterances of
temporally sensitive sentence-types? He should start with a clause
for atomic sentence-types on which clauses for complex sentences
involving tense operators can be built recursively. The obvious
place to look for such a clause is the model theory of tense logics.
Deleting the reference to a model, for an atomic sentence, X—'it is
raining,' say—this will include some such clause as:
(1) For any time t, truet(X) iff it is raining at t,
and the clause
(2) For any time t, and any sentence S, true t ('P'(5 )) iff there is a
time t', earlier than t such that 5 is true/
for the past-tense operator 'P.' Can these clauses be interpreted by
the Priorean presentist so as to yield his own account of the truth
conditions of token utterances of such sentence-types? Given the
practice we have followed hitherto of identifying the 'correct'
employment of a sentence-type with the truth of a token utterance,
this question engages the discussion in Evans (1979). For Evans
(1979) rightly ponders the significance such clauses have for the
'correctness or incorrectness,' or 'semantic values,' of token
utterances. As he says, any semantic theory worth its salt must
combine with knowledge of the world to guide the production and
assessment of token utterances.
The most radical interpretation of these clauses Evans considers,
which he labels "T\,y connects 'true t ! with the correctness/semantic
value of token utterances via the claim
(6)(S)(u)(t)[Of(S,u)-»-(Correct-at-t(u)^truet(S))]
Evans then observes that the effect of (6) is that clauses such as (1)
and (2) are interpreted in such a way that
All utterances of the same type have the same semantic value ...
[but] an utterance is not to receive a single assessment as correct
or incorrect, but rather an assessment which varies with time ...
the evaluation of an utterance as correct or incorrect depends

108
 A Presentist's Refutation of Mellor's McTaggart

upon the time the evaluation is made (and so the evaluation

varies) (p. 348)
I hope this remark seems familiar! Once token truth is identified
with token correctness, Evans's characterization of the consequence
of using (6) to interpret (1) and (2) coincides with the consequence
we have already seen to be drawn from the supposition that e.g.
'Bush is sitting down' is non-indexical with respect to time in
Mellor's (1986, 1998) presentation of his dilemma's first horn. This
is the consequence that all tokens of this sentence-type have the
same truth value whatever their temporal location, and this truth
value switches between true and false as time ('the time of
evaluation') progresses and Bush goes about his daily business.
Moreover, like Mellor, Evans rejects this consequence, and hence
Tj. Unlike Mellor, however, he does not protest at the supposition
that all tokens of a sentence-type like 'Bush is sitting down' have the
same truth value whatever their temporal location. He merely
protests that (p. 349):

If a theory of reference permits a subject to deduce merely that a

particular utterance is now correct, but later will be incorrect, it
cannot assist the subject in deciding what to say, nor in
interpreting the remarks of others.

Even more striking, from our point of view, is the fact that Evans
discerns Tj, and hence this unpalatable consequence, in Prior's
metaphysics, and in particular in Prior's sympathy for the
scholastic doctrine about propositions (above). Evans's point, of
course, is that this doctrine appears likewise to warrant the
conclusion that token utterances of e.g. 'Socrates is sitting' do not
admit a stable evaluation as correct or incorrect (nor, hence, as true
or false), but instead switch between correct and incorrect (and
hence between true and false) with the passage of time depending
on whether Socrates is sitting. All that seems needed to mediate the
passage from the scholastic doctrine about propositions to this
conclusion about token utterances is the attractive assumption that
(*) For all token utterances u, the truth condition of u is met if
and only if the proposition it expresses is true
In short, Evans (1979), which Mellor has never acknowledged,
anticipates even the later, supplemented version of the first horn of
Mellor's (1986, 1998) semantic development of McTaggart's
argument.

109
Philip Percival

III.in An alternative presentist semantics—Notwithstanding the

attractiveness of Evans's thought that Prior's scholastic doctrine of
propositions entails radical token truth value variability (between
truth and falsity) over time, I no longer think it has this
consequence. On a presentist reading (*) does not combine with the
scholastic doctrine about propositions to give switches in the truth
value of a token utterance between true and false, while combining
(6) with clauses like (1) and (2) does not yield the sort of semantic
clauses that Prior's presentism demands. Combining (6) with (1)
and (2) yields semantic clauses such as the following:
(6*) (u)(t)[Of('Bush is President of the U.S.,' u)->
(True-at-t(u)-^Bush is President of the U.S. at t)]
However, clauses such as these are not at all in keeping with Prior's
philosophy of time. It is no more in keeping with that philosophy to
specify the truth conditions of token utterances of sentence-types
involving tenses like 'it will be the case that' by quantifying over
times than it is in the spirit of a modalism according to which there
are primitive modal operators to explain the truth conditions of 'OS'
by quantifying over possible worlds. Admittedly, quantification
over times does play a fundamental role in the model-theoretic
semantics for tense-logics which Prior pursued. Ultimately, however,
I believe his metaphysics obliges him to think of such model theories
as no more than heuristic devices.21 Far more natural, in the light of
the scholastic doctrine of propositions, is a clause like this:
(T) (u)[Of('Bush is President of the U.S.,'u)->(True(u)-^Bush is
President of the U.S.)]
Clauses like this seem to be the explicit target of the first horn of
Mellor's dilemma, and they seem an implicit target of Evans's
anticipation of it. For such clauses give all token utterances of a
given temporally sensitive sentence-type the same truth conditions.
Accordingly, this clause seems to support the absurd consequence
the two of them detect in the supposition that e.g. 'Bush is President
of the U.S.' is non-indexical with respect to time. That (T) ascribes
all token utterances of this sentence-type the same truth conditions,
and, hence, the same truth value, appears to conflict with the
assumption that being temporally sensitive, the correct use of this
sentence-type, and hence the truth value of a token utterance of it,
depends on the token's temporal location. Moreover, although Bush
is President of the U.S., he soon won't be. And this fact seems to
21
See Percival (1990).
110
 A Presentist's Refutation of Mellor's McTaggart

combine with (T) to yield the consequence that while token

utterances of 'Bush is President of the U.S.' are true, they will be
false.
Nevertheless, these objections to (T) miss the point of Prior's
metaphysics. I will address them in turn.

The argument within time—For Prior the quantifier in (T) does not
range over tokens having different temporal locations. According to
Prior, only the present is real. Part of what this comes to is that the
first-order universal quantifier ranges over only presently existing
entities. In that case (T) does not assign tokens of 'Bush is President
of the U.S.' having different temporal locations the same truth
conditions, and hence the same truth values. It only assigns truth
conditions to the existing, present tokens over which its quantifier
ranges.
On the other hand, there is an obvious objection to the suggestion
that reading (T) this way deflects the argument 'within' time. An
argument to the effect that a weak reading of (T) has only weak
consequences is beside the point. On such a reading, (T) is too weak
to serve as a semantic theory! A semantic theory for a natural
language must capture a speaker's grasp of the truth conditions of
token utterances, and speakers do know the truth conditions of a
token utterance of e.g. 'Bush is President of the U.S.' which is not
located in the present once the temporal location of the token is
given to them. I know not only what the truth conditions of a pre-
sent utterance of that sentence-type are; I also know what the truth
conditions of an utterance tomorrow of it will be, and what the
truth conditions of a past utterance of it were. On Prior's reading of
its quantifier, (T) does not capture this knowledge. The fact that (T)
survives the argument within time is therefore irrelevant.
However, while I endorse the viewpoint which underlies this
objection, I do not accept that it shows that (T) can only achieve the
generality semantic theory demands if its quantifier is construed as
ranging over non-present tokens, thereby leaving it vulnerable to
the argument over time. It is not the case that in order to capture a
speaker's knowledge of the truth conditions of non-present token
utterances a semantic theory must quantify over token utterances
other than present ones. This is readily seen by reflecting on the
modal analogue. My grasp of English equips me to understand
actual token utterances of the sentence-types of English. In
particular, given relevant information regarding the context of
utterance, I can assign truth conditions to actual token utterances.
But I can do more: I can understand equally well, and assign truth
111
Philip Percival

conditions to, token utterances that are non-actual. No one

addressed to me yesterday the words 'y° u a r e wearing a blue shirt.'
However, in virtue of my grasp of English, I know that had anyone
addressed those words to me, their utterance would have been true
if and only if I had then been wearing a blue shirt. Here then is
semantic knowledge which appears to go beyond knowledge of the
truth conditions of actual utterances. Is this to say that semantic
theory cannot proceed by quantifying over all actual token
utterances and assigning truth conditions to them; that it must
quantify over non-actual token utterances too? Unless the
metaphysics of modality is much more straightforward than is
generally believed, it is not to say this! On one respectable view,
there are no non-actual objects. On this view, far from being
obligatory, quantifying over non-actual token utterances so as to
capture a speaker's grasp of the truth conditions of non-actual
token utterances is not an option. Rather, a speaker's grasp of this
much is captured by embedding quantifiers which actually range
over actual objects in a necessity-operator thus:
Necessarily, (u) [Of('Bush is President of the U.S. on Christmas
Day 2002,' u)->(True(u)<->Bush is President of the U.S. on
Christmas Day 2002)]
Since Prior assimilates time to modality as conceived by the modalist,
a Priorean semantics for the tenses will treat the temporal case
similarly. Speakers of English know the truth conditions of non-
present token utterances, and a semantics for English must capture
this knowledge. But this is not to say that such a semantics is
obliged to quantify over non-present token utterances. On the
contrary, from the point of view of Prior's presentism, so doing is
not even an option: there are no such utterances. Rather, grasp of
the truth conditions of non-present token utterances is captured by
embedding the clause which quantifies over present token
utterances in a temporal operator, thereby strengthening (T) to:
(TT) Always, (u) [Of('Bush is President of the U.S.,' u)->
(True(u)<-+Bush is President of the U.S.)]
Using this clause, I can capture what I know about the truth
conditions e.g. of yesterday's utterances. What needs to be captured
is the knowledge that a token utterance yesterday of 'Bush is
President of the U.S.' was true iff Bush was President of the U.S.
yesterday. On the view that I am recommending, this knowledge
just is knowledge of a sort of instance of (TT), namely:

112
 A Presentist's Refutation of Mellor's McTaggart

(TTY) Yesterday, (u) [Of('Bush is President of the U.S.,' u)->

(True(u)<->Bush is President of the U.S.)]
Suppose that I know that yesterday, someone uttered 'Bush is
President of the U.S.' I can put this knowledge together with my
knowledge of English and knowledge of current affairs to deduce
that this utterance was true, i.e. that yesterday, someone produced a
true token utterance. For we have
Yesterday, (x)(Fx->\Gx-<->p); Yesterday, 3xFx;
Yesterday, p | - Yesterday, 3x(Fx&Gx)
(TT) does consistently what Mellor says cannot be consistently
done: it assigns all token utterances of a temporally sensitive
sentence-type like 'Bush is President of the U.S.' the same tensed
truth conditions, and it does so while respecting the undeniable
semantic fact that the truth condition, and hence the truth value, of
a token utterance of such sentence-types depends on the temporal
location of the token.
As originally presented, then, the first horn of Mellor's (1981)
McTaggart doesn't even engage (TT) and the doctrine that only
what is present exists. Hence, it begs the question against Prior;
hence it begs the question against the tensed view of time. But what
about the additional argument 'over' time by which Mellor (1986,
1998) supplements the first horn?

The argument over time—As we have seen, Le Poidevin (1991)

effectively concedes that presentism can accommodate the
argument 'within' time while insisting that it is refuted by the
argument 'over' time.22 But Le Poidevin's evaluation of this
supplementary argument is wrong. A further consideration exposes
it as a simple fallacy. Suppose Bush is President of the U.S., and
that Tomorrow(Bush is not President of the U.S.). One cannot use
(TT) to obtain directly from this the consequence that all token
utterances of 'Bush is President of the U.S.' will have a truth value
tomorrow different from the truth value they have now: one needs
the assumption that they will tomorrow be utterances of 'Bush is
President of the U.S.' Admittedly, for any token utterance of 'Bush
is President of the U.S.' for which it holds that tomorrow, it will be
(a token) of 'Bush is President of the U.S.,' we can derive both that
this utterance is true (from T), and that it will be false (from TT).
However, the assumption that token utterances persist over time
isn't forced upon the presentist as a matter of logic, and nor do I
22
See above, footnote 15.
113
Philip Percival

think he should accept it. Certainly, e.g. marks on a page do persist

through time in this way. But a charge simply that presentism has
the consequence that the truth values of certain sentence tokens
switch between true and false with the passage of time is vacuous.
As Mellor insists, that they do so is undeniable on any view. Take an
unattended electronic clock which has stopped and which continues
to display the words 'it is now 17.00 hours.' During the temporal
interval in which it persists, this display switches from being false to
being (briefly) true every twenty four hours. Similarly, a shop sign
saying 'Closed' left forgotten in a shop window switches its truth
value as the shop opens and closes.23 Accordingly, whether or not the
truth value of a sentence token of a sentence-type like 'Bush is
President of the U.S.' switches its truth value in this way during the
temporal interval in which it persists is not what is at issue. What is
at issue, rather, is whether it switches its truth value at times at
which it does not persist.
This issue is best addressed by focusing not upon token
sentences, but upon token utterances. Token utterances are the
primary objects of semantic evaluation. Though they are often
effected by producing sentence tokens, they are nevertheless
distinct from them. If I write the words 'I love you' on a piece of
paper and push it through a grating on a prison visit, the primary
object of evaluation—my utterance—is not identical to that
inscription. If three years later, the inmate to whom these words are
addressed takes a crumpled piece of paper from her pocket and
thinks 'no you don't, not any more,' she is not questioning the
sincerity of my utterance, three years previously, whatever the
consequences of her remark for the token of the sentence-type 'I
love you' which she holds in her hand.
What is at issue is whether or not presentism is committed to the
doctrine that the truth value of a token utterance of a sentence-type
like 'Bush is President of the U.S.' switches between true and false
with the passage of time. (TT)—namely
Always, (u) [Of('Bush is President of the U.S.,'
u)->-(True(u)<->Bush is President of the U.S.)]

23
Cf. Mellor's (1981: 99) remark that 'long-lasting thing tokens can vary
in truth-value during their lifetimes—e.g. a token of "e is past" printed
before e will start off false and end up true.' (Actually, I think Mellor's
treatment of this sort of case is unduly simplistic: once made, the
distinctions between 'true' and 'true in L,' and between 'thing' token and
token utterance, suggest a subtler treatment. (See Percival (1994: 203—5).)
Nothing hangs on these subtleties.)
114
 A Presentist's Refutation of Mellor's McTaggart

—has this commitment if token utterances exist over time. But for
the presentist they don't. Everyone can agree that they don't persist
over time. Token utterances are acts which, like winning the lottery,
are instantaneous. The verbs which describe them are 'success
verbs.' But to equate existence with present existence, as the pre-
sentist does, is to say that entities exist only while they persist. So
the fact that token utterances don't persist over time will be taken
by the presentist to indicate that they do not exist over time. For the
presentist, just as it cannot be the case that Bush is sitting down iff
Candorice is swimming except when Bush and Candorice exist, and
hence except at times during which they persist, neither can it be
the case that some token representation r is true iff Bush is sitting
down except when r exists, and hence at times at which r persists. It
follows that (TT) does not commit the presentist to supposing that
the truth value of a token utterance of e.g. 'Bush is sitting down'
switches between true and false over time depending on Bush's
actions. On the contrary, such an utterance has a truth value only at
times at which it exists, and hence at times at which it persists.
Therefore, since such an utterance is momentary, it has only one
truth value.

IV Conclusion

Mellor's McTaggart, and in particular its first horn, only appears

forceful if doctrines which are central to Prior's presentism—namely,
the related doctrines that only present objects exist and that only
present objects fall within the range of the first-order quantifier—
are ignored. Whatever (the temporal variability of) the semantic
properties of persisting tokens of sentence-types like 'Bush is
President of the U.S.,' the fundamental objects of semantic
evaluation, and, hence, the fundamental bearers of the 'correctness'
predicate (which I have read as a truth-predicate), are not tokens of
sentence-types, but token utterances of sentence-types. These are
more or less instantaneous. For the presentist, to suppose that fifty
years ago there existed a token of 'Bush is President of the U.S.' is
not to suppose that some token utterance of this type exists and has
certain truth conditions. It is to suppose that: fifty years ago (there
is something which has certain truth conditions). My knowledge of
English includes implicit knowledge of what those truth conditions
were. For I know that: fifty years ago, for all u, [Of('Bush is
President of the U.S.,' u)-KTrue(u)<->-Bush is President of the
U.S.). Knowing a little history, I can work out that: fifty years ago,

115
Philip Percival

all utterances of 'Bush is President of the U.S.' are false. This does
not contradict the fact that: all utterances of 'Bush is President of
the U.S.' are true. Nor does it have the consequence that there is
some token representation which has truth values at times at which
it does not persist. Nor, hence, does it have the consequence that
there are cases of temporal variability in token truth value other
than the trivialities of the kind Mellor acknowledges, such as a
token of a temporally sensitive sentence-type having different truth
values at different times at which it persists. Prior's presentism is
therefore untouched by Mellor's McTaggart.
Is this slight of hand? I've exchanged Mellor's official definition
of the 'tenseless' view for something stronger, thereby weakening
the tensed view. Moreover, I've couched his semantic version of
McTaggart's argument against the A-series in terms of a sentence,
'Bush is sitting down,' which does not include a predicate which
appears to ascribe pastness, presentness or futurity. Haven't I
changed the goalposts, and thereby missed the point?
Not at all. These manoeuvres were conducted on Mellor's behalf.
To see this, consider Mellor's own example, 'e is past.' Mellor
claims that it is contradictory to suppose that all tokens of this
sentence-type are made true by the one A-fact that e is past: not
only would so doing give all tokens of this sentence-type the same
truth value, contrary to the fact that this sentence-type is what I
have called 'temporally sensitive', the truth value which all tokens
of it are given at any one time would also switch over time, e.g. from
false to true in the case in which e starts off future and becomes
present and then past, contrary to the fact that token truth value
does not change in this way. However, it is easy to see that these
claims beg the question against the presentist semantics I have
sketched on behalf of Prior. In keeping with (TT) we have: Always,
(u)[Of('e is past,' u)->(True(u)<->e is past)]. On the assumption that
it is a fact that e is future, this truth condition has the consequence
that all tokens of 'e is past' are false. However, this consequence
does not engage entities which merely will or did exist, and, hence,
which will fall or which fell within the scope of the quantifier in the
future or in the past. A fortiori, contrary to the first horn of
Mellor's dilemma as originally presented—what I have called the
argument 'within' time—it does not engage any merely future
tokens of 'e is past' that Mellor says are true. To be sure, on the
assumption that e.g. it is a fact that two days hence (e is past), it will
also follow that two days hence (all tokens of 'e is past' are true).
Again, however, this further consequence only engages those exist-
ing tokens of 'e is past' which, e.g. being scrawled on a blackboard,
116
 A Presentist's Refutation of Mellor's McTaggart

persist for at least two days. This small subset of tokens will indeed
be true two days hence, even though they are false now. But this
consequence is not absurd. (At any rate, it is a consequence which
Mellor himself believes.) By contrast, since existing token
utterances of 'e is past,' and other similarly short lived tokens of it,
will not persist for the next two days, the fact that two days hence
(all tokens of 'e is past' are true) says nothing about them. A fortiori,
contrary to the argument by which the first horn of Mellor's
original dilemma is eventually supplemented—what I have called
the argument 'over' time—it does not say about them that two days
hence they will have a truth value different from the one they have
now. Therefore, on a presentist semantics, the supposition that
token utterances of a sentence like 'e is past' have tensed truth
conditions is not contradictory. It has neither the consequence that
some token utterance of this sentence-type is both true and false
(pace the argument 'within' time), nor that some token utterance of
this sentence-type which is false will become true at some later time
at which it does not persist (pace the argument 'over' time).
Accordingly, Mellor's semantic development of McTaggart's
argument against the A-series fails to engage Prior's presentist
metaphysics. The tensed view of time survives it unscathed.24

References

Bigelow, J. 1991. 'Worlds enough for time,' Nous 25, 1-19.

1996. 'Presentism and properties,' Philosophical Perspectives 10,
35-52.
Dolev, Y. 2000. 'Dummett's anti-realism and time,' European Journal of
Philosophy 8, 253-76.
Dummett, M. 1960. 'A defence of McTaggart's proof of the unreality of
time,' pp. 351-57 of M. Dummett (1978).
1969. 'The reality of the past', pp. 358-74 of M. Dummett (1978).
1978. Truth and Other Enigmas, London, Duckworth.
1981. Frege: Philosophy of Language, 2nd edition, London,
Duckworth.
Evans, G. 1979. 'Does tense-logic rest upon a mistake?' pp. 343-63 of his
Collected Papers, Clarendon Press, Oxford.
Geach, P. 1949. Critical Notice of Julius Weinberg, Nicolaus of Autricourt,
Mind 58, 238-45.
24
I would like to thank Yuri Balashov, Craig Bourne, Jim Edwards,
Jonathan Lowe, and Nathan Oaklander for helpful comments. I am
grateful to the Arts and Humanities Research Board for funding research
leave during which this paper was completed.
117
Philip Percival

Lowe, E. J. 1987a. 'The indexical fallacy in McTaggart's proof of the

unreality of time,' Mind 96, 62-70.
1987b. 'Reply to Le Poidevin and Mellor,' Mind 96, 539^2.
1992. 'McTaggart's paradox revisited,' Mind 101, 323-26.
1998. 'Tense and persistence,' pp. 43-59 of R. Le Poidevin (1998).
McTaggart, J. M. E. 1927. The Nature of Existence, Cambridge,
Cambridge University Press.
Oaklander, L. N. and Smith Q. (eds) (1994), The New Theory of Time,
Yale University Press, New Haven.
Poidevin, R. Le and Mellor, D. H. 1987. 'Time, Change, and the
"Indexical Fallacy",' Mind 96, 534-38.
1991. Change, Cause and Contradiction, Clarendon Press, Oxford.
1998. Questions of time and tense, Clarendon Press, Oxford.
Mellor, D. H. 1981. Real Time, Cambridge University Press, Cambridge.
1986. 'Tense's tenseless truth conditions,' Analysis 46, 167-72.
1988. 'I and now,' Proceedings of the Aristotelian Society 89, 79-94.
1998. Real Time II, Cambridge University Press, Cambridge.
Paul, L. A. 1997. 'Truth conditions of tensed sentence-types,' Synthese
111,53-71.
Percival, P. 1989. 'Indices of Truth and Temporal Propositions,'
Philosophical Quarterly 39, 190-97.
1990. 'Indices of truth and intensional operators', Theoria 57,
148-72.
1991. 'Knowability, actuality, and the metaphysics of context-depen-
dence,' Australasian Journal of Philosophy 69, 82-97.
1992. 'Thank goodness that's non-actual,' Philosophical Papers 21,
191-213.
1994. 'Absolute truth,' Proceedings of the Aristotelian Society 94,
189-213.
Priest, G. 1986. 'Tense and truth conditions,' Analysis 46, 162-7.
Prior, A. N. 1967. Past, Present, Future, Clarendon Press, Oxford.
1970. 'The notion of the present,' Studium Generate 23, 245-8.
Smith, Q. 1994. 'The truth conditions of tensed sentences,' pp. 69-76 of
L. N. Oaklander and Q. Smith (eds) (1994).

118
Time and Degrees of Existence:
A Theory of 'Degree Presentism'
QUENTIN SMITH

1. Introduction
It seems intuitively obvious that what I am doing right now is more
real than what I did just one second ago, and it seems intuitively
obvious that what I did just one second ago is more real than what
I did forty years ago. And yet, remarkably, every philosopher of
time today, except for the author, denies this obvious fact about real-
ity. What went wrong? How could philosophers get so far away from
what is the most experientially evident fact about reality?
The concept of a degree of existence (of being more or less real)
went out of fashion with the rise of analytic philosophy early in the
20th century, specifically, with Russell's 1905 article 'On Denoting',
for in 1904 and earlier years he and G. E. Moore held a sort of
Meinongian theory of degrees of existence (subsistence and exis-
tence are distinguished, with existence being a higher degree of
being than subsistence). Early work by Frege also rejected the
notion of degreed existence and implied that existence is an all or
nothing affair; either something exists or it does not exist, and it
makes no sense to talk about it existing to some degree.
Most (but not all) philosophers from Plato to Meinong have held
doctrines of degrees of existence. Unfortunately, however, they also
denied this obvious temporal fact about reality, for they explained
degrees of reality in other ways than the way we know it (as being
more or less distant from the present). Indeed, they typically held
(at least from Plato to Hegel and Bradley) that a being that does not
exist in time at all is what is most real. Time, they often said, is
unreal. Philosophy has been and still is a flight from temporal real-
ity. There are a large number of reasons why philosophers have
denied the obvious nature of reality, most of them being logically
independent of one another. It would take a book to discuss all these
reasons, and so I shall instead concentrate in this essay on explain-
ing and defending the logical coherence the most obvious of all
experientially obvious facts.
Being temporally present is the highest degree of existence. Being
past and being future by a merely infinitesimal amount is the second
119
Quentin Smith

highest degree of existence. Being past by one hour and being future
by one hour are lower degrees of existence, and being past by 5 bil-
lion years and being future by 5 billion years are still lower degrees of
existence. The degree to which an item exists is proportional to its
temporal distance from the present; the present, which has zero-
temporal distance from the present, has the highest (logically) pos-
sible degree of existence.
These degrees are quantifiable in terms of their opposites,
degrees of nonexistence. The present has a zero degree of non-
existence. What is one second past has a one second degree of
nonexistence, and what is two seconds past has a greater degree of
nonexistence, namely, a two second degree of nonexistence.
There is a difference of degree and not of kind between the pre-
sent and what is no longer present or not yet present. This is shown
by the fact that our present mental state includes temporal parts that
are past by 1/millionth of a second, etc., and this small degree of
pastness is such a high degree of existence that we cannot experien-
tially distinguish it from present existence, 100% existence. These
degrees of existence are immediately given in our phenomenological
experience. I believe this theory is logically coherent unless one
misinterprets it by assigning a different meaning to 'degree of exis-
tence' or 'degree of nonexistence' than I have assigned it.

2. The Types of Tensed Theories of Time

Philosophers of time today are either tensed theorists or tenseless

theorists. The theory of degrees of existence is a type of tensed
theory of time, what is most accurately called (in today's parlance)
'degree presentism', to indicate that every item is distanced from
the present to some degree (amount of time). This is the first time
the phrase 'degree presentism' has been used. But I can think of no
more accurate name for this theory. The A-theory or tensed theory
of time can be divided into five kinds (at least):

i. The three-dimensional equal reality theory. The tensed theory

McTaggart articulated is such a theory; the theory he articulated
(but did not endorse) implied that future events, present events and
past events are equally real. Richard Gale and George Schelesinger
held different versions of the three-dimensional equal reality
theory. They differed from McTaggart's formulation in several
ways, however; for example, they argued B-relations are analysable
into A-properties. The problem with this theory is that the past no
120
 Time and Degrees of Existence

longer exists, whereas the present does exist, and this entails the
present has a higher existential status than the past, not an equal
status.

ii. C. D. Broad held a two-dimensional equal reality theory. He held

that present events and past events are equally real, but that the
future is nothingness, i.e. that it is not the case that there is a future.
Broad avoided answering such questions as: why am I preparing a
lecture for tomorrow if there is no future? And when I expect the
mailperson to arrive in the room in the next few minutes, what am
I expecting if it is not the case there is anything I am expecting?
And what are weather reports about? Are they about nothingness
itself? If so, maybe Heidegger was on to something. Craig [2000a;
2002] holds that future tense sentence-tokens corresponds to
presently existing, abstract states of affairs. This seems problematic
since most future tense sentence-tokens (e.g., 'the sun will explode
in 5 billion years') are about concrete things or events, not abstract
objects. Further, all of these sentence-tokens are not about some-
thing that is wholly in the present and only in the present, but some-
thing that is not yet present. Craig would be better off if he either
claimed that no future tensed sentence-token is true (it is without
truth value or has the value of false) or else reductively analysed the
future tense into something nontemporal, such as modals, as does
Ludlow [1999].
What is distinctive about these two or three dimensional equal
reality theories is that present events do not have greater reality than
past and future events or that present events are the only real events.
Presentness is neither identical with existence nor logically equiva-
lent to existence, since events exist regardless of whether they are
past or present (e.g., on Broad's theory). For example, Broad writes:
'There is no such thing as ceasing to exist; what has become exists
henceforth for ever.' [Broad, p. 88]. However, even apart from the
problems I have briefly mentioned above, it is beyond imagining
what my dying is if it is not my ceasing to exist, so Broad's philos-
ophy at the very least needs some elaboration if it is not to seem
manifestly false.
The equal realities are versions of the standard contemporary 'all
or nothing' theory of existence; i.e., either a particular exists or it
does not exist, and there is no in between state where it exists to
some degree and does not exist to some degree.

iii. Degree presentism. This is the theory of temporal degrees of

existence that I shall defend in this essay. This theory is a presen-
121
Quentin Smith

tism since it holds that what exists in the maximal or perfect degree
of existence is only what is present. It is degreed since it holds that
the past and future are not wholly unreal, but are real to some less
than maximal degree. When I remember and expect things, I am
not remembering and expecting nothing at all. The past and future
exist to some degree, but to a lesser degree than the present. To say
that Socrates exists to some degree does not mean, for example, that
as he recedes into the past he first loses a hand, then a leg, then his
head, etc.; what it means is that as he recedes into the past his dis-
tance from the present increases, e.g., from being 2,400 years from
the present to being 2,4001 years from the present. I am not merely
stipulating that 'degree of existence' means 'distance from the pre-
sent'; I claim that this is how we experience existence, as something
with degrees, and thus that degree of existence = distance from the
present accurately describes our immediate acquaintance with
existence and time. If you deny that this is how you experience exis-
tence, it is consistent for me to explain this fact by saying that 'you
are in the grip of a (false) theory' and this prevents you from recog-
nizing experienced existence and time. Alternatively, I could take on
the larger project of arguing at length that every other theory of
time and existence is either logically invalid or empirically discon-
firmed. But here I am taking the first step of arguing that degree
presentism is a logically consistent theory, and thus that the 'hand
wave dismal' of the concept of degrees of existence that began with
Russell's 1904 'On Denoting' has just as much argumentative sup-
port that Russell gave his 'all or nothing' theory in his 1904 article
(i.e., no support at all—he 'refuted' Meinong by 'calling him
names', e.g., by saying he 'lacks a robust sense of reality'.) It is not
without interest that 20th century theories of existence have their
original 'justification' in an insult.
Thus, I do not agree with Tooley [1997: 233] when he writes
about 'the position that Quentin Smith refers to as 'presentism' in
his book Language and Time (New York: Oxford University Press,
1993). But Smith's usage seems very unfortunate, since presentism
so understood, is compatible with the existence of past states of
affairs See, esp., p. 165'. Tooley is correcting in citing the specific
page where this view is expressed in some form. However, each
sense of 'exists' I distinguish on this page gives presentness a max-
imal existential status and pastness and futurity a lower status, since
the past or future exist in a nonmaximal sense, viz., I say on page
165 that they 'exist' in the sense that they are no longer present or
are not yet present. What Tooley, Zimmerman, Craig, Ludlow,
Markosian, Bigelow and others refer to as 'presentism' is only one
122
 Time and Degrees of Existence

version of presentism, namely, the solipsistic version, where pre-

sentness not only has the maximal existential status but the only
existential status of any sort or degree whatsoever.

iv. Modal, solipsistic presentism. Only the present exists in any

meaningful sense of 'exists' and it is the not the case that any past
or future event or thing exists in any sense whatsoever. 'Modal' is
used in the possible world sense, since the present is conceived as
analogous to the actual world and the past and future to merely pos-
sible worlds. Prior originated this view and William Craig (2000a;
2002] developed the ontology of this view to the greatest extent.
Most tensed theorists of time hold this view in recent times: it is
held by Christensen, Lloyd, Levison, Wolterstorff, Chisholm,
Zimmerman, Markosian, Bigelow and others. According to this
theory, you have no past or future, since it is not the case that there
is a past and future. I believe solipsistic presentism is logically self-
contradictory. The main founder of solipsistic presentism, Prior,
tellingly defines it in an implicitly self-contradictory way, a way
endorsed by Craig, Zimmerman and other solipsistic presentists.
Prior writes: '...the present simply is the real considered in relation
to two particular species of unreality, namely the past and the
future.' [Prior, 1998: p. 80]. If the real stands in relation to the unreal,
the unreal is real, since only something real can stand in relation to
something. Unreality can no more stand in relations than it can pos-
sess monadic properties. If one says that Prior means that a thinker
is considering the present in relation to unreality, then my response
is that the consideration is self-contradictory, since I cannot consis-
tently consider the unreal to stand in a relation to the real. Further,
there can only be multiple species of real things; unreality cannot be
differentiated into 'two particular species', as Prior says. It is an
implicit contradiction suppose that there is some differentia that
differentiates one sort of nothingness from another sort of nothing-
ness, since no differentia exist in nothingness or nonexistence.
(More precisely, it is a contradiction that 'if everything is in the pre-
sent, then something is in what is not present, viz., differentia and
species'.) If one is a presentist, one is forced to be a degree
presentist on pain of holding a logically self-contradictory theory
(solipsistic presentism).
Notwithstanding this, Craig should be commended for recogniz-
ing some of the logical incoherencies that largely make up Prior's
schematic ontology for solipsistic presentism; Craig begins by quot-
ing Prior's statement '. . .the reality of the present consists in what
the reality of anything consists in, namely the absence of a qualify-
123
Quentin Smith

ing prefix'.' In the next sentence Craig comments on Prior's quot-

ed remark. 'This last remark [of Prior] illustrates the sort of con-
flation of semantics and ontology that so exasperates Smith, for the
reality of the lamp before me on my desk does not consist in the
absence of a qualifying prefix, since prefixes do not operate on
lamps.' [Craig, 2000a: pp. 193-194]'. Degree presentism does not
face such difficulties.

v. There is reductivist solipsistic presentism, a recent and novel

view first articulated by Peter Ludlow in [1999], where he reduces
the past and future tenses by proposing that they fall instead into
the linguistic category of evidentials (for the past tense), which have
evidence for the proposition expressed as their semantic relata; and
modals (for the future tenses), which have dispositions as their
semantic content. This theory seems promising, since it avoids the
problem of how irreducibly past tensed sentence-tokens can be true
if there is no past, or how irreducibly future tensed sentence-tokens
can be true if there is no future. But this reductivist theory has
problems of coherency of its own. For example, it has to overcome
such objections as that 'Some space existed for which there is no
evidence', which is contingently true or false, if reductively
analysable in terms of evidentials, becomes the self-contradiction
that there is evidence for some space for which there is no evidence.
And 'A new spatial point p will come into existence in the future,
even though nothing present has the disposition for this point's
existence' is contingently true or false but is self-contradictory if it
means that something present has the disposition for a spatial point p's
existence, even though nothing present has the disposition for this point's
existence.

By a tensed theory of time I mean a theory that takes tensed truths

and tensed states of affairs to be basic. This way of defining a
'tensed theory of time' makes Michael Tooley's theory [1997] a
tenseless theory of a dynamic time. Whether it is 'dynamic' in some
intelligible sense in which other tenseless theories are not dynamic
is a debatable question [Smith, 2000].
I have a limited goal in this essay; I aim merely to argue that
degree presentism is logically unproblematic and is thereby theoret-
ically preferable to modal solipsistic presentism. I discuss this
modal solipsistic presentism since it is the most prevalent tensed
theory of time and because it is, in my opinion, the most logically
incoherent theory of time.
I make no attempt to answer critics of the tensed theory by tense-
124
 Time and Degrees of Existence

less theorists such as Oaklander [1996], Dyke [forthcoming],

Graham Nerlich [1998] and D. H. Mellor [1998]. This would
require several papers unto themselves. This paper is degree
presentism versus modal solipsistic presentism.

3. Degree Presentism, Tenseless Exemplification, and

Existence

Ironically, one of the main arguments of solipsistic presentism is

that degree presentism is logically incoherent, whereas degree pre-
sentism holds that solipsist presentism is incoherent. One thesis of
degree presentism is that there is no primitive, irreducible tenseless
quantifier; there is no tenseless sense of 'exists' that cannot be
analysed into more basic tensed senses of 'exists'. I did not develop
a 'degrees of existence' theory in my 1993 book, Language and
Time, but I still hold the view I state on page 165: Language and
Time, namely that 'x exists' in the tenseless sense means 'x existed,
exists or will exist', where the middle 'exists' is present tensed. This
shows that some interpreters of Language and Time, such as
William Craig, are wrong in imputing to me the doctrine that every
event exists equally in a primitive, irreducible tenseless sense of
'exists'. Such a view is the way McTaggart conceived of the tensed
theory of time, but I reject the idea that properties of futurity, pre-
sentness and pastness successively inhere in events that exist in an
irreducible tenseless sense of 'exist'. Solipsistic presentists do not
distinguish between the equal reality tensed theory of time, such as
the one put forth by McTaggart or Gale, and the degrees of exis-
tence theory. But this is not to impute a misunderstanding of some
text to these solipsists, for the degrees of temporal existence theory
has not been formulated before, and so when I examine their criti-
cal comments, this is primarily for heuristic purposes, not to show
that they have misunderstood some doctrine that neither I nor any-
body has stated in some book, such as Language and Time.
Solipsistic presentists deny not merely that there are properties of
pastness and futurity, but even (in some cases) that there is a prop-
erty of presentness. A. N. Prior said '. . . the presentness of an event
is just the event. The presentness of my lecturing, for instance, is
just my lecturing.' [Prior, 1998: p. 81]. But this cannot be true, for
if the presentness of event E is wholly identical with E, then 'E is
present' means E is E. But it is not a tautology that E is present, but
contingently true or false, whereas it is a tautological truth that E is
E.
125
Quentin Smith

But what most troubles solipsistic presentists is the idea that past
and future things and events possess properties. If they are past or
future, they do not exist (i.e., are not present) and thus there is
'nothing there' to possess any properties, even properties of being
past or future. In response, I think that this is the point where a
degrees of existence theory can be introduced to clarify the
apparent problems that nonpresent items possess properties.
Equally troubling to many philosophers is my earlier claim that
nonpresent items presently possess properties. I held this view in
Language and Time, and Oaklander [1996], Craig [2000a: pp.
189-217] and Zimmerman [1998: pp. 212] have all strongly object-
ed to this theory. At first I thought they were wrong, but now I
think they are on the right track (even if I don't agree with the
details of their criticisms of Language and Time). This doesn't mean
I accept the solipsistic presentism of Craig and Zimmerman or
Oaklander's tenseless theory of time. Rather, I prefer to respond by
developing a new version of the tensed theory of time, degree
presentism.
I would first note that monadic predicates (predicates are linguis-
tic items) of past and future events are abbreviations of relational
predicates, for a nonmaximal degree of temporal existence requires
every determination of a particular to be a relationship to the pre-
sent, in relation to which the degree to which the past or future
particular exists is determined. For example, Socrates does not
presently have the nonrelational property of being alive. Nor does he
presently have the relational property of having been alive over 2,000
years earlier than the present time. Socrates existed 2,000 years ago,
so he cannot exemplify relational properties in the present. But this
past Socrates can stand in relations to the present of being earlier
than it.
Whatever had been F, had been F, not timelessly, not at the past
time at which it was F (for at that time, the thing is (present tense)
F rather than had been F), and not at the future time. Having been
alive is analysable into the property of aliveness and the state S of
the thing tenselessly being alive being earlier than the present time.
It is the whole complex, the state S, that stands in this relation to
the present, not the thing's tenseless exemplification of being alive.
'Pastness', 'was', 'have been', 'had been' and the like are analysable
into the exemplification of the property F that the thing possessed
at the time it was present, and the complex state S consisting of
thing's exemplification of this property being related to the present
time by the relation of being earlier than it. Here exemplification
can be taken to have a primitive tenseless meaning (that is fine, since
126
 Time and Degrees of Existence

I do not identify exemplification with existence). If it were tensed,

we would have to ask if the thing's exemplification is past, present
or future, and this would lead to an unpalatable infinite regress (the
exemplification is present, and the exemplification of presentness is
present, and so on ad infinitum), as Oaklander pointed out very
insightfully in his [1996]). Socrates' having been alive is analysable
into tenselessly exemplifying the property of aliveness, such that
the state of Socrates' tenselessly exemplify this property is over
2,000 years earlier than the present.
This means, contrary to my theory in Language and Time, that
Socrates does not presently possess the property of having been
alive. It is not the case that Socrates lies in the past but that his
EXEMPLIFICATION of having been alive lies in the present.
Rather, according to my new theory of degrees of existence, the
semantic content of 'having been' is that Socrates exemplifies
(tenselessly) the relational property being alive over 2,000 years ear-
lier than the present time. Socrates is past and his exemplification of
properties is a tenseless 'tie' (to use Strawson's term) of the proper-
ties to Socrates, such that the tenseless exemplification is atemporal
in the sense that it has no A-properties and stands in no B-relations.
N-adic property-ties are not the kind of item that is tied to other n-
adic properties. Property-ties are not monadic properties or
relations, but are what 'ties' properties and relations to entities; in
the more usual terminology, property-ties are not properties or
relations but are things' exemplifications of properties and things'
standings in relation. Since these 'ties' do not have A-properties or
stand in B-relations, they are 'atemporal' in this sense, but they are
'temporal' in the sense that the property-ties belong to a state that
has temporal n-adic properties
The complication of the tenses still preserves this relatedness to
the present. For example, if I say that Thales had been dead before
Socrates was born, we have two past tense expressions, each of
whose semantic content includes a relation of being earlier than the
present time. The state S composed of Thales' being (tenselessly)
dead is earlier than the present time and is earlier than the state S'
composed of Socrates' birth; in addition, the state consisting of
Socrates' being born is earlier than the present.
The present is existence itself, ipsum esse. As many philosophers
have suggested, existence does not neatly fall into any category of
what exists. Existence is not a thing, event, property, relation, set,
mathematical object, proposition, operator, and so on. It is unique.
Since existence is the present, the same holds for the present. We
may say that each maximal existent is a presence (something present)
127
Quentin Smith

and that the whole of maximal existents is The Presence (or, if you
prefer, the present, or the present time). But these are primitive
notions, just as Plantinga says possibility, actuality and necessity are
primitive notions. You can understand them by examples or
synonymous expressions, but they cannot be defined in terms of
something else. (Craig [2000a] says existence is the 'act of exempli-
fication' but since acts need subjects, 'act' is at best a metaphor,
since nobody performs the act of exemplification (Craig presum-
ably has in mind a deity). And existence cannot be exemplification,
since 'exemplification exists' is a contingent truth and yet 'exempli-
fication is an exemplification' is a logically necessary truth, and does
not even imply that exemplification exists. The present tense sense
of exists is the 'is' (present tense) in 'x is', or is 'a presence' in 'x is
(tenselessly) identical with a presence'. Maximal existence is also
conveyed in 'x is (tenselessly) simultaneous with the present' and 'x
is (tenselessly) a part of The Presence'. Why should such basic
notions as exists and presence need to be defined in order to be
understood? They don't, since our ability to find false definitions of
exists (present tense) and presence presupposes that we already
understand the meaning of 'exists' and 'presence'. A maximal exis-
tent is a presence. (Craig accurately notes that this was my first
theory of existence, in my 1986 book The Felt Meanings of the
World. To Craig's credit, he said [2000a] that existence would be
presentness if it were not for the fact that some things can exist
timelessly. But the credit is only partial. Why should the possibility
of timeless existence bar the identity existence with presentness? A
timeless existent is a presence that (a) occupies only one instant, this
instant being the present, and (b) belongs to a possible world in
which there is only that instant—and thus no past or future
instants.)
Note that my theory of degrees of existence implies that all pred-
ications are reducible to tenseless predications, involving only a
tenseless copula 'is' or verb phrase (e.g., 'runs'). Every relation to
the present, xRy, where y is the present, is such that x stands tense-
lessly in the relation R to the present. Since we have eliminated
monadic properties of pastness and futurity, we need only one irre-
ducibly tensed word to state our ontology, namely, the present
tensed 'exists'. Actually, we can go further and have our entire
ontology stated in tenseless language, for we can replace 'exists' by
'is (tenselessly) simultaneous with the present' and 'the present' is a
noun phrase, whereas tense is (by definition) an adverbial modifica-
tion of a verb.
But this does not mean we belong in Tooley's camp. For Tooley,
128
 Time and Degrees of Existence

there are only tenseless facts. For me, there are only tensed facts
(where 'tensed' now has the nonlinguistic, ontological sense of A-
facts, as distinct from B-facts). Every fact includes a relationship to
the present. This is why I call my theory a presentism, or, more
fully, degree presentism.
What is present stands tenselessly in a relation of simultaneity to
the present. For example, I am (tenselessly) simultaneous with the
present. What of the question, when am I simultaneous with the
present? This question is malformed, since the information about
the temporal location of myself is already given in the question
itself. If I am tenselessly simultaneous with the present, my tempo-
ral location (by definition) is the present time. The sentence 'the sky
is blue' means that the sky tenselessly exemplifies blueness and the
state S composed of the sky's blueness is simultaneous with the pre-
sent. Blueness is a nonrelational property of the sky, but the state of
the sky being blue stands in a relation of simultaneity to the present.
The fact that past and future individuals lack nonrelational prop-
erties reflects their ontological status as not fully real beings; in a
sense, they are partial beings. Does this mean they have another part
that is nonbeing? It seems absurd to say that something is partly a
being and partly a nonbeing. I respond that this sentence can be
interpreted in many different ways, and most of these ways result in
the sentence being taken to express a self-contradictory proposition.
But there is a consistent way to interpret it. The sentence
'Socrates is partly a nonbeing or a nonexistent' means two things (a)
he has no nonrelational properties, and (b) he lacks full being of the
amount, 2, 400 years (to use an approximate date), which means he
is temporally separated from the full being by 2,400 years.
The sentence 'Socrates is partly a being or existent' means (a) he
has only relational properties and (b) he partakes of full being in the
sense that he is not present but tenselessly stands in certain metric
relation to the present, a relation of being distant from it to a cer-
tain amount. (I am a realist, not a conventionalist, about time's met-
ric, so something's degree of existence is not a matter of an arbi-
trary convention.)

4. Does Degree Presentism Imply Past and Future Particulars

are Nearly Bare Particulars?

Zimmerman objects to presentist theories that imply realism about

the past and future. He criticizes the relevant parts of Language and
Time by avowing that 'A painful headache cannot exist without
129
Quentin Smith

being painful ... Plato cannot exist while having neither body nor
soul. What's left of these past and future things and events is too
thin... Neither Plato nor the headache has any of these ordinary
intrinsic properties it displayed while present... Past and future
things become nearly-bare particulars' [Zimmerman, 1998: p. 212].
Let us see if we can isolate the structure of this argument against
degree presentism about the past or future. First, let us clear up
Zimmerman's fallacies of equivocation upon 'exists' and 'is' before
we get to the heart of the matter.
The quoted sentences can be reformulated in a coherent way by a
degree presentist. Does degree presentism imply that a painful
headache is not painful? No. A degree presentist would agree with
everyone else that the statement 'a painful headache is not painful'
is an explicit logical contradiction. What would a degree presentist
say about painful headaches? He would say that a headache had
been painful while it existed, but since the headache has passed
away it is not now paining anyone. To derive a contradiction from
degree presentism, we need to equivocate on tensed expressions.
Notice that by saying 'a painful headache' the tense is omitted, so
we do not know from this expression when this headache occurs—
whether it is past, present or future. This is tantamount to treating
it as tenselessly existing; we (or, rather, Zimmerman)imply it exists,
but are omitting to say whether it existed, exists or will exist. Now
if we say 'a painful headache is not painful' this conversationally
implicates (in Grice's sense) that the 'is' is used tenselessly, since we
are predicating a property of an event that we have identified as
existing tenselessly. It is a clear contradiction to say, using the tense-
less 'is' in the irreducible B-sense, that 'a painful headache is (tense-
lessly) not painful'.
But suppose we do not use misleading language and fallacies of
equivocation to describe the theory of degrees of existence. Then
we would say that the headache, although painful while it was pre-
sent, is not now paining me, and it is not now paining me precisely
because it is no longer present.
And certainly the degree presentist believes that Plato cannot
exist without having a body and soul. This means that Plato cannot
be present without at the same time having a body and mind. And
it implies that if Plato had been present, then Plato had a body and
mind while he had been present. But it certainly does not mean that
Plato has an irreducible tenseless existence and lacks a body and
mind while he tenselessly exists. And it certainly does not mean that
Plato is present and presently has no body and mind. And it does not
mean that when Plato was present, he lacked a body and soul. It is
130
 Time and Degrees of Existence

true that Plato is tenselessly earlier than the present by over 2,000
years, and his having a body and mind is tenselessly earlier than the
present by exactly the same amount of time. But statements of this
sort are supposed to be where the problem with the degrees of exis-
tence theory lie. Where is the problem?
I think with this new theory of degrees of existence I have
avoided a main problem that Zimmerman, Oaklander and Craig
have noted with the theory in Language and Time. I there held that
past things presently exemplify monadic properties. This implies
that the things are not present, but their exemplification of proper-
ties lies in the present. How could a thing lie in the past and its
states lie in the present? Zimmerman, Oaklander and Craig are
right; I should abandon this theory. Contra Craig, and more in line
with Oaklander, this theory is not logically self-contradictory but is
an implausible synthetic assertion; it may be considered as a syn-
thetic a priori falsehood. Even if it not a synthetic a priori falsehood,
but merely implausible, it seems preferable to adopt a more intu-
itively plausible theory. The degrees of existence theory implies
that no nonpresent items presently exemplify properties. Rather
past or future items tenselessly stand in relations to the present of
being earlier than it to a certain degree or later than it to a certain
degree.
Zimmerman, Oaklander and Craig will undoubtedly have some-
thing to say about whether or not this new theory is 'better' than the
old theory, since Oaklander rejects the tensed theory of time and
Zimmerman and Craig reject degree presentism. But for now, let us
be sure we really have in fact solved the above-discussed problem
that Zimmerman posed for any theory that the past and future are
real in some sense.
Is there a contradiction in the degrees of existence theory I for-
mulated?
Note there is no logical contradiction in the statement:
(1) x is no longer present but x tenselessly stands in relation to the
present of being earlier than it to a certain degree.
A contradiction would be 'x wholly is no longer present and x wholly
is present' or 'x tenselessly stands in relation to the present of being
distant from it to a certain degree and x does not tenselessly stands
in relation to the present of being distant from it to a certain
degree'. But these contradictions and other contradictions cannot
be derived from statement (1).
Let us focus on the distinction between past particulars that the
solipsistic presentist calls 'nearly-bare particulars' and present par-
131
Quentin Smith

ticulars, which are 'fully clothed' and thus seem ontologically

unproblematic to the solipsist. It is true that past particulars lack
the 'ordinary intrinsic properties they display while present'. In
what sense does this make past particulars nearly bare particulars in
any ontologically problematic sense?
The unusual feature of degree presentism is summarized as this:
Past (or future) particulars do not have nonrelational, monadic
properties, but only stand in relations or have relational properties.
Thus they are 'bare particulars' in the sense that they lack nonrela-
tional, monadic properties. This 'bareness' is due the fact that these
particulars are only partly real; they are partly unreal in the sense
(among other senses) that they are bare in this respect.
The property of being past is, when ontologically analysed, a
relational property. If something is past, it is past by two hours, or
past by 7 minutes, etc. Past particulars are partly clothed in the
sense that they have relations in which they stand to the present of
being temporally distant from it to some degree (amount of time).
'Plato walked' means Plato tenselessly exemplifies walking over
2,000 earlier than the present. This temporal distance from the pre-
sent is another sense is which past particular are partly unreal, for
the present is existence, full reality, and past particulars acquire only
a degree of existence by virtue of being earlier than the present, by
virtue of standing in relation to existence of lacking existence by a
partial amount of it (e.g. the amount, 2,000 years).
So we have this result: maximal existents have nonrelational
monadic properties and also stand in relations. But particulars that
exist to less than the maximal degree only stand in relations. This is
one sense in which they are partly real and partly unreal. Let us ask
ourselves again; does it involve a logical contradiction?
I believe it can be proven not to be a contradiction. For any pre-
sent item x, and for each nonrelational or relational property F than
a present item x can possess, x has F or x does not have F. This is
the precise meaning of the phrase 'the present item x is a logically
complete individual', i.e. satisfies the logically necessary criteria to
exist in the tenseless sense (existed, exists or will exist).
Past items also meet this criterion. For any past item y, and for
each nonrelational or relational property F than a past item y can
possess, y has F or does not have F. The past item cannot have any
nonrelational properties, and so it is does not have any such proper-
ties as being spatial, being mental, breathing, and the like. For each
nonrelational property G, it lacks G. But for each relational property
R, it either has R or lacks R. For each nonrelational property G it
possessed when it was present, it possesses the relational property of
132
 Time and Degrees of Existence

having possessed G a certain amount of time ago. The past partic-

ular is bare of nonrelational properties, but this is a necessary con-
dition of it being past to some degree; but it is clothed to a logically
sufficient degree. That is, it meets all the logically necessary condi-
tions to exist in a tenseless sense (to have existed, to exist or to exist
in the future). This implies that the past item x is a logically com-
plete individual.
But how can a particular exist to any degree without having the
nonrelational property of being a particular? The answer is that it
has a relational property of being a particular. It had been a partic-
ular (say) 150 years earlier than the present time, but it is false that
it is (present tense) a particular. There is no such particular that
occupies the present time. And since the particular tenselessly
exemplifies its relational properties of being temporally distant
from the present, none of the states of the particular are present.
This enables us to answer the problem Zimmerman formulated:
He writes that the degree doctrine implies the following: 'Plato is
still a substance, I suppose, but he doesn't talk or think or walk or
sleep or have any spatial location' [Zimmerman, 1998: p. 212]. Now
we can see two problems. First, does the degree presentist hold that
Plato is still a substance, that is, is presently a substance? This, of
course, would pose a problem, for it would then be the case that
Plato is presently a substance but presently is not in space and, fur-
ther, presently has no mind.
But no such substance occupies the present time. It is not the case
that Plato is still a substance. The nonrelational property of being a
substance is not presently possessed by Plato. Rather, Plato had been
a substance while he existed, over 2000 years distant from the pre-
sent time. Substantiality, like every other property possessed by
Plato, characterizes Plato only in the sense that it is part of a relation
Plato has to the present. The state of Plato's tenseless exemplification
of being a substance is over 2,000 years earlier than the present.
What is it that is earlier than the present? It is not an existent. But
if it is not an existent, how can it stand in a relation to an existent?
The answer is that it does exist—to some degree. To say that 'x does
not exist', where 'exist' means 'present', can be analysed as meaning
x does not exist to the maximal degree but exists to a less than max-
imal degree. What is the particular that is receding from the pre-
sent? It is neither a total nonbeing nor a total being. It is a partial
existent, which is part way between total nonbeing and total being.
Its partial nonbeing consists in its lack of nonrelational properties
and its lack of full existence. The 'part of being' it lacks is identical
with the interval of time that separates the being from the present
133
Quentin Smith

(the present having complete, whole or maximal being). This theory

is not absurd unless one attaches different senses to the terms 'par-
tial being', 'degrees of existence', 'maximal existence' than I have
given them in this essay. These phrases may have emotional associ-
ations with Bradley, Hegel, Aquinas, Plotinus, Plato and others, but
that is not a problem with my theory, but a problem with your emo-
tions.
This seems to be a plausible way to explain our phenomenologi-
cal experience of time; for we do experience that uttering the begin-
ning of this sentence is more real than A Cro-Magnon's grunt
35,000 years ago. And we experience tomorrow's visit to the dentist
as more real than the day of our 85th birthday twenty or forty years
from now.
Thus degree presentism is, in a sense, half-way between
solipsistic presentism and the tenseless theory of time. Degree pre-
sentism denies that the past and future are nothingness (distin-
guishing it from solipsist presentism) and denies that the past and
future are equally as real as the present (distinguishing it from the
tenseless theory of time, as well as from the equal reality version of
the tensed theory of time).
The problem raised by the solipsist for the degree presentist
about nonexistents is a question-begging problem of the solipsist's
own making. The solipsistic presentist assumes that existence is 'all
or nothing' and then infers from this that there is nothing earlier
than the present that could stand in any relation to the present. But
this is tantamount to assuming at the outset of the debate that solip-
sistic presentism is true and degree presentism is false.
Semantics issue about reference remain to be discussed. The
name Plato' refers to a maximal existent when used while Plato is
present, but the name 'Plato' refers to a lesser degree existent, when
Plato is no longer present. It is the same particular that is the refer-
ent, but a referent that exists to different degrees at different times.
What I said above needs to be made more precise. I said Plato is
a particular. But is he? The answer is that Plato's substantiality, par-
ticularity and thinghood are only partly real, since they are over
2,000 years distant from what is wholly real, what is present. Plato
had been a substance, had been a particular, had been something,
over 2,000 years ago. Plato is a particular to a certain degree, name-
ly, a degree that is over 2,000 years from the time when Plato's par-
ticularity was maximally existent. Is this unintelligible? No. It just
means that Plato is (tenselessly) a particular 2, 400 years before the
present time.
But there may be other problems for the degrees of existence
134
 Time and Degrees of Existence

theory, leading to logical contradictions. The solipsistic presentist

may say that a particular essentially has nonrelational properties
when it is present, but loses these essential properties when it
becomes past. How can a particular lose an essential property? The
answer is that the property is essential to the particular in the sense
that the particular cannot be present without possessing the prop-
erty. The particular is essentially a human, and being human is an
essentially nonrelational property. When the particular becomes
past, it possesses an essentially relational property, one that is the
past-time version of the presently possessed property. Instead of it
being true that x has the essentially nonrelational property of being
human, it is now true that x tenselessly has the essentially relation-
al property of having been a human over 100 years earlier than the
present.
The problem of change may be pressed further. How can a par-
ticular change relational properties over time if the particular is not
present? I see no problem here. As each second passes, a past par-
ticular loses one relational property and acquires a new relational
property of being one second more remote from the present time.
What about other properties? Consider Plato existing or being
present when 389 B.C.E. is present. William Craig [2002] believes
the realist runs into problems here. When this time is present, Plato
possesses the property of being alive. But how can Plato, as located
in 389 B.C.E., also possess the property of being dead? How can
Plato as located in 389 B.C.E be both alive and dead? First let us
remove the fallacy of equivocation on 'is'. In the premise the 'is' is
used in a tensed sense and in the conclusion it is used in a tenseless
sense. The proper way to state this fact is that Plato as located in 389
B.C.E. was alive when 389 B.C. E. was present, but Plato as located
in 389 B.C.E. is now dead since 389 B.C.E. is over 2,000 years
earlier than the present.
Once this equivocation is removed, we can understand how Plato
as located in 389 B. C. E. can change properties. The answer is triv-
ial: By the passage of time. When Plato, as located in 389 B.C.E., is
over 2,000 years earlier than the present time, then Plato-in-389
B.C.E. does not presently possess the property of being alive but
instead tenselessly possesses the relational property of being dead
for over 2,000 years. Plato-in-389 B.C.E. possesses the property of
being alive when 389 B.C. E. is present, and does not possess this
property when 2000 A.D. is present. Thus, one cannot deduce the
contradiction that Plato-in-389 B.C.E. simultaneously possesses
logically incompatible properties. (A more precise treatment of this
issue could be given if we gave two analyses, one presupposing the
135
Quentin Smith

continuant theory of particulars and the second presupposing the

temporal parts theory of particulars. But that is not necessary here.)
In conclusion, I think the intuitively plausible degrees of exis-
tence theory can be defended with respect to its logical coherency
and solves more ontological conundrums than does modal, solipsist
presentism.

References

Broad, C. D. 1998. (see The Big Questions)

Craig, William Lane. 2000a. The Tensed Theory of Time. Dordrecht.
Kluwer Academic.
2000b. The Tenseless Theory of Time. Dordrect. Kluwer Academic.
2002. 'Presentism: A Defense' in Quentin Smith and Alex Jokic (eds),
Time, Tense and Reference. Cambridge, MA. MIT Press.
Oaklander, Nathan L. 1996. 'Smith and McTaggart's Paradox'. Synthese.
Smith, Quentin 1993. Language and Time. New York. Oxford University
Press.
1986. The Felt Meanings of the World: A Metaphysics of Feeling. West
Lafayette, Ind. Purdue University Press.
Van Inwagen, Peter and Dean Zimmerman. (1998), The Big Questions.
Cornwall. Blackwell Publishers.
Prior, A. N. 1998. (see The Big Questions)
Zimmerman, Dean 1998. (see The Big Questions).

136
McTaggart and the Truth about Time
HEATHER DYKE

1. Introduction
McTaggart famously argued that time is unreal. Today, almost no
one agrees with his conclusion.' But his argument remains the locus
classicus for both the A-theory and the B-theory of time. I want to
show how McTaggart's argument provided the impetus for both of
these opposing views of the nature of time. I will also present and
defend what I take to be the correct view of the nature of time.
McTaggart begins by noting that, when we think about when, in
the temporal order of things, an event is located, there are two ways
in which we can do this. On the one hand, we can locate an event as
in either the past, the present, or the future. Once we have desig-
nated an event as occurring, say, three days ago, then every other
event temporally related to that event will have some determinate
location in either the past, the present, or the future. McTaggart
called the series of events ordered in this way the A-series. But we
can also locate events in time without reference to the past, present
or future. We can locate events as temporally related to each other.
We say that an event is earlier than, later than, or simultaneous with
some other event. We can use these relations to order every event in
a temporal series. McTaggart called the series of events generated
in this way the B-series.
This claim of McTaggart's is an uncontroversial one about the
ways in which, as a matter of fact, we think about the temporal loca-
tions and ordering of events in time. The A-series and the B-series
are just two different ways of ordering the very same events and
moments. For instance, the Great Exhibition of 1851 occupies an
A-series location: it is 149 years in the past. It is also located in the
B-series. It is, for example, 63 years earlier than the outbreak of
World War I, which implies nothing about its location in the past,
present, or future. By drawing this distinction between the A-series
and the B-series, McTaggart has simply drawn our attention to the
fact that we can represent the temporal ordering of events in these
two different ways. But in the light of this distinction, genuinely
substantial metaphysical questions arise: is one of these two ways of

' Sprigge (1992) is an exception.

137
Heather Dyke

representing the temporal ordering of events more fundamental

than the other? Does one of them truly represent the nature of
time?
One characteristic of the A-series, that the B-series lacks, is that
events don't keep the same A-series position for very long. If an
event is present, then very soon it will be past. An event that is
already past is gradually becoming more past. The B-series, on the
other hand, is what we might call a static ordering of events. If an
event occurs two days earlier than another event, then those two
events are forever related to each other in that way. So, the notion of
the A-series involves what we might call A-series change, which has
also been called the flow of time, or temporal becoming.
Having made this distinction between the A-series and the B-
series, McTaggart proceeds to present his argument for the unreal-
ity of time. It consists of two theses: a positive and a negative the-
sis. The positive thesis is that, if time exists at all, it must involve an
A-series. His argument for this depends on the claim that there
could not be change unless the events and moments of time formed
an A-series as well as a B-series. So, the A-series is essential for
there to be change, and change is essential for there to be time. His
negative thesis is that the notion of the A-series is self-contradictory,
so it cannot be part of reality. The conclusion that McTaggart
draws is that, since the A-series must exist if there is to be time, but
the A-series cannot exist because it is a self-contradictory notion,
time itself does not exist.
In general, philosophers have accepted one of McTaggart's
theses and rejected the other. So, while they recognize that his
argument is valid, they have thought it unsound. However, they
have disagreed over which thesis to accept and which to reject. The
A-theorists agree with his positive thesis, that the A-series is
essential for the existence of time. A-theorists think that a descrip-
tion of time that does not make reference to the A-series is an
incomplete description of temporal reality. Consequently, A-
theorists reject McTaggart's negative thesis, that the notion of the
A-series is self-contradictory. B-theorists, on the other hand, tend to
accept McTaggart's negative thesis. The notion of the A-series is
indeed self-contradictory, so the A-series cannot be part of reality.
But they reject his positive thesis. They think that time can exist
without its constituents forming an A-series. In particular, they
argue that change is possible without the elements of time occupy-
ing A-series locations. So, the A-series is self-contradictory, but
since it is not essential to time, time itself is real, but consists only
of a B-series.
138
 McTaggart and the Truth about Time

2. Why I reject McTaggart's positive thesis

McTaggart thinks that change is of the essence of time. There is a

sense in which we all think this, since we all think that time is the
dimension of change. Change occurs when something possesses
incompatible properties at different times: a tree is fully clothed
with leaves, and then bare, and then fully clothed once more. But
McTaggart means something more than this. For him, time itself
exhibits change. Times, and the events that occur at them, change
from being future to being present to being past. When McTaggart
claims that time is the dimension of change, he means that it is the
dimension of A-series change.
Why does McTaggart think that the existence of change requires
the existence of an A-series? He argues that if time consisted only
of a B-series, change would not be possible. If all there is to time is
B-series facts about the temporal relations between events, then
there cannot be change, according to McTaggart, because B-series
facts never change. Facts about the B-series relations between
events are fixed; they do not change. The only way in which the
characteristics of an event can change is if it changes from being
future to being present to being past. McTaggart's charge against
the B-theory can be put another way. If there is only a B-series so
that all events are equally real, no matter when they occur, and no
event ever changes its B-series location, then nothing really changes.
Reality is a fixed and unchanging entity.
McTaggart's objections to a B-series account of change are, I
submit, question-begging. He argues that nothing about a B-series
ever changes, so the B-series cannot accommodate change.
However, he assumes, for the sake of his argument, that change
means A-series change. It may be true that the B-series itself never
changes, but that doesn't mean that the constituents of a B-series
cannot undergo change. It may be true, to use McTaggart's exam-
ple, that if a poker is hot at one time and cool at a later time, noth-
ing about those facts ever changes, but it doesn't follow that those
facts do not constitute a change in the poker. McTaggart is assum-
ing that the paradigm subjects of change are events. It is events that
change from future to present to past. But a proponent of B-series
change need not accept this assumption. She can argue instead that
the paradigm subjects of change are objects. It is objects that
change by having incompatible properties at different times.
To put my objection in another way, McTaggart's argument
establishes nothing more than that without an A-series there cannot
be A-series change. A B-theorist can accept this, because for her,
139
Heather Dyke

there is no A-series, and there is no A-series change. McTaggart's

conclusion is a conditional. He claims to have established that if
there is time, then there must be A-series change. But all that he has
really established is that if there is an A-series, then there must be
A-series change. This conditional is acceptable to a B-theorist, since
for her it is true because both antecedent and consequent are false.
If the existence of time depends on the existence of A-series
change, then it would indeed follow that without an A-series there
could not be time. But all McTaggart has established is that the
existence of the A-series depends on the existence of A-series
change.

3. Why I accept McTaggart's negative thesis

McTaggart's argument that the notion of the A-series involves a

contradiction is deceptively simple, and strangely uncompelling on
a first reading. His premises are that the A-series positions are
incompatible, and that if the A-series exists, and with it A-series
change, then every event occupies every A-series position. It follows
from these premises that the A-series does not exist. The obvious
response, as McTaggart notes, is that no event satisfies all of the
incompatible A-series predicates at the same time, but only succes-
sively, and there is no contradiction in anything satisfying incom-
patible predicates at different times.
The obvious response, however, doesn't work. It says that noth-
ing is ever past, present and future at once, but only at different
times. There are two ways in which we can cash out this response.
Are the different times at which an event is past, present and future,
different times in the A-series or in the B-series? Taking the second
option first, the response now goes as follows: of course nothing can
be future, present and past. But something can be future at one
time, <i, present at a later time, t2, and past at a still later time, £3.
This way of understanding the obvious response does indeed avoid
the contradiction. But it is unacceptable to anyone wishing to retain
a genuine A-series in her ontology. To say that an event, E, is future
at tj, present at t2, and past at £3, is to say no more than that E occurs
at t2, which is later than t± and earlier than t3. The A-series claims
collapse into B-series claims. By anchoring the possession of incom-
patible properties to different times in the B-series, the A-series,
and A-series change, have fallen out of the picture. Qualifying the
A-series claims in this way yields B-series claims, which do not
change their truth-value as things change their A-series position.
140
 McTaggart and the Truth about Time

All that is described by these qualified claims is a fixed and

unchanging B-series.
So in order to avoid the contradiction, and retain A-series facts
and change, a defender of the A-series must relativize the posses-
sion of the incompatible A-series predicates to different times in the
A-series. Now the response goes as follows: of course nothing can
be future, present and past. But something can be present now,
while it was future and will be past. This move also succeeds in
removing the contradiction, but it does so by introducing a set of
second level temporal predicates, and while some of these are com-
patible with each other, there are some that are not. But if the A-
series, and A-series change, are real, then every event possesses
every second level temporal predicate, even the incompatible ones.
So the contradiction has not been removed, merely shifted up to
these second level temporal predicates.
What has happened is this. By saying that an event is present, was
future, and will be past, we have described things as they are now.
But because reality undergoes A-series change, things have not
always been as they are now, and they won't remain as they are now.
In order to incorporate A-series change into our description of A-
.series-involving temporal reality, we must recognize that the same
event also will be future, and was past. But these second-level tem-
poral predicates are incompatible with the ones we used to avoid the
contradiction in the first place. So, relativizing the possession of
incompatible A-series predicates to different times in the A-series
cannot eliminate the contradiction.

4. An alternative expression of McTaggarts paradox

I often find that people are initially resistant to McTaggart's rea-
soning in establishing his negative thesis. I therefore wish to
unearth the contradiction that he identified in a different way.
Recall McTaggart's A-series. The properties of events by which
they are ordered in the A-series are the properties of being past,
being present and being future.2 If we suppose that events really are
ordered according to these A-series characteristics, then we must
also admit that they change their A-series characteristics over time.
An event, like the Sydney Olympics, was once in the remote future,
and was recently in the near future. It is now in the present, will
soon be in the recent past, and will gradually recede into the more
2
There are also finer gradations of A-series locations such as being three
weeks ago, being this week, and being two minutes hence.
141
Heather Dyke

remote past. The question for McTaggart, and for us, is: does time,
in reality, exhibit these A-series characteristics? Do events really
possess the characteristics of being in the past, present or future,
and do they really change in respect of them over time?
Let's suppose that events really do possess these characteristics.
In that case, we can plausibly suppose that they are properties in
some sense. Indeed, this is a common way of ascribing an ontolog-
ical status to these characteristics, by those who think they are real.3
How do the properties of pastness, presentness and futurity differ
from each other? One thing that we can say is that they have differ-
ent extensions. The property of pastness applies to all those things
that are earlier than the present moment. Presentness applies to all
those things that are occurring simultaneously with the present
moment. Futurity applies to all those things that occur later than
the present moment. But now notice that I have presented a picture
of temporal reality that is only accurate for a moment. We can dis-
tinguish between past, present and future, in terms of their exten-
sions, but by doing so, we leave out the other feature of the A-series:
the continual change from future to present to past that everything
undergoes.
So we must try to distinguish between past, present and future in
a way that accommodates A-series change. But accommodating A-
series change removes our means of distinguishing between past,
present and future. Because everything successively possesses every
A-series property, it follows that the extensions of the properties of
pastness, presentness and futurity are all exactly the same. They all
apply to everything. And it is not simply that these properties have
the same extensions as a matter of mere contingent fact. If A-series
change occurs, then they necessarily have the same extensions, rein-
forcing the conclusion that there is no genuine distinction between
them.
One could object that the extensions of these properties are not
identical if there is a first or a last moment of time. A first moment
of time is never future, and a last moment of time is never past. But
this does not avoid the co-extensiveness objection. In that picture,
the property of presentness is co-extensive with the property of
being either past or future, or alternatively, of being non-present.
My conclusion stands, as being present, and being non-present are
necessarily co-extensive if A-series change occurs.
Here, then, is McTaggart's paradox in my terms. We can only
distinguish between the properties of pastness, presentness and
3
See for example, Smith (1993).
142
 McTaggart and the Truth about Time

futurity at some moment of time. But this yields a static 'snap-shot'

picture of tensed time, a picture that is patently false, because
everything is constantly changing its A-series property. But as soon
as we try to incorporate A-series change into the picture, we lose our
means of distinguishing between the A-properties. The distinction
between pastness, presentness and futurity collapses because every-
thing successively possesses them all.
To put my point another way, to suppose that the A-series is real
requires commitment to two theses. Firstly, one must hold that
there is a real, observer-independent distinction between past, pre-
sent and future. Secondly, one must hold that different distributions
of past, present and future obtain at different times. But it seems
that one cannot hold both of these theses. Marking the objective
distinction between past, present and future requires leaving A-
series change out of one's account because one can only distinguish
between past, present and future at a particular moment of time.
Holding the second thesis, that the distribution of pastness, pre-
sentness and futurity changes from moment to moment, involves
relinquishing our grip on the first thesis, that there is an objective
distinction between past, present and future. As the distribution
between A-properties changes the distinction between them col-
lapses, since they all apply to everything. The entire account thus
collapses under the weight of this contradiction.
It follows that time cannot be such that its constituents form an
A-series. To suppose that the A-series is real is to suppose that time
has these two features: an absolute distinction between past, present
and future, and a continual change in respect of this distinction that
the constituents of time undergo. But time cannot possess both of
these features, so the A-series is not real.

5. The A-theory's options

If, as I have argued, McTaggart's attack on the reality of the A-

series succeeds, what options remain? One option is simply to deny
that times and their contents form an A-series at all. There is no
objective, observer-independent distinction between past, present
and future; nothing really changes from being future to being pre-
sent to being past. Taking this line involves explaining why, if there
is no past, present, and future, we are misled by our experience into
thinking that there is. But for many this sort of response will be
unsatisfactory. What is needed, they argue, is not an error theory of
our possession of A-series concepts, but an account of them that
143
Heather Dyke

does not collapse in the face of McTaggart's paradox. I think there

are two potentially viable options for those sympathetic to the A-
theory, which I will briefly outline.
McTaggart himself suggests the first option when he says 'It is
never true that [an event] is present, past, and future. It is present,
will be past, and has been future.' (McTaggart (1927) 21). For
McTaggart the explanation cannot stop here, since it merely intro-
duces more complex tenses than the three simple ones, and because
A-series change is continually occurring, every event has every
complex tense, just as it has every simple tense, and some of them
are incompatible. So, as we have seen, for McTaggart this line of
response cannot avoid the contradiction. But an A-theorist could
take issue with McTaggart's claim that the explanation cannot stop
at this point. Take any event, E, that is happening now. We can say
of E that it is present, was future, and will be past, and in saying this
we do not contradict ourselves. What we have described is the pre-
sent state of affairs. E has the property of being present. It also has
the past tense property of being future, and the future tense prop-
erty of being past. Provided the A-theorist is willing to concede that
the present state of affairs is all that there is, she can avoid
McTaggart's paradox.
There are some A-theorists, presentists, who are willing to make
this concession,4 and I grant that it does offer a way out of
McTaggart's paradox, but whether it can constitute a viable meta-
physics of temporal reality is another question. For many, commit-
ment to the unreality of past and future will be too high a price to
pay for avoidance of McTaggart's paradox. Those presentists who
are willing to pay it must still show us that their picture of the world
is coherent, and coheres with our experience. And it is not obvious
that they can do this. The presentist's response to McTaggart effec-
tively involves denying that A-series change takes place. Certainly
the presentist can talk about events that will be present, and events
that will, in a week's time, be two weeks past, and this way of talk-
ing gives the impression that A-series change is consistent with the
presentist picture. But all these expressions really convey is that
every event is located somewhere in the A-series, and that, were a
different moment present, they would be located elsewhere in the
A-series. Presentism, it seems to me, cannot accommodate the
change in A-series positions that events and times undergo, for as
soon as it attempts to do so, it falls right back into McTaggart's
paradox.
4
For example, Prior (1970), Bigelow (1996), Craig (1998), and Hinchliff
(1996).
144
 McTaggart and the Truth about Time

The second A-theoretic response to McTaggart, that seems

viable at first sight, is suggested by Dummett (1960). Dummett
argues that what McTaggart's argument really shows is that there
cannot be a complete description of reality independent of some
perspective. According to Dummett, McTaggart implicitly assumes
that there can be such a description. When that assumption is com-
bined with his thesis that the temporal cannot be completely
described without the use of A-series expressions, the contradiction
quickly follows. If time is real, then the complete description of
reality contains incompatible facts, viz., for any event E, E is past,
present and future. McTaggart concludes that, since the complete
description of reality cannot contain incompatible facts, time is not
real. Dummett concludes instead that the false premise is the one
that says that there can in principle be a complete description of
reality. So time is real, but reality only contains some of the incom-
patible temporal facts.
If there can be no complete, observer-independent description of
temporal reality, then one of two possibilities follows. Either tem-
poral reality consists of two domains: that which we can consistently
describe and that to which we can in principle have no epistemic
access. If this is the right interpretation of Dummett, then the
burden of proof lies squarely with him. Why should we think that
there is any more to temporal reality than that to which we have
epistemic access? Alternatively, we can interpret Dummett as argu-
ing that the maximal consistent description of temporal reality that
can be given from a particular temporal perspective describes all
that there is. This alternative reduces to presentism. We can give a
complete description of the A-series location of every event given a
particular temporal perspective, and this would constitute a com-
plete description of present fact. If present fact is all there is, then
presentism is true. But if this is the right interpretation of
Dummett, then he faces the same problem that I outlined above for
the presentist.

6. Moving on from McTaggart

I think that time itself exists, but that the A-series doesn't. There
are no characteristics of pastness, presentness or futurity. There is
no flow of time. Nothing really changes from future to present to
past. But all of our temporal experience seems to suggest that there
is an A-series. How come we seem to be deceived by our experience
on such a massive scale? In what follows I will present an account
145
Heather Dyke

of our temporal experience that appeals only to the existence of a B-

series.
The feature of our experience that is most suggestive of the exis-
tence of an A-series is that we talk about events as if they were
located somewhere in the past, present or future. We say, for exam-
ple, 'It's nearly 5 o'clock', which suggests that 5 o'clock is located in
the proximate future, and will soon be present. We say 'World War
II ended 55 years ago', which suggests that the end of World War
II is located 55 years in the past. And when we say things like this,
what we say is determinately either true or false. I will call sentences
like this, which appear to locate events or times somewhere in the A-
series, A-sentences. It is undeniable that many A-sentences are true
when they are uttered, but what makes them true, if not the fact that
a certain event or time is located somewhere in the A-series?
According to the B-theory of time, the fact that makes a sentence
like 'World War II ended 55 years ago' true is the fact that the event
that it is about (the end of World War II) is 55 years earlier than the
utterance of the sentence.5 All events stand in fixed and unchanging
B-series relations to each other. Utterances of sentences are events
like any other, so they stand in temporal relations to other events. In
particular, they stand in temporal relations to the events that they
are about. An A-sentence that appears to locate an event somewhere
in the A-series will be true if and only if that event and the utter-
ance of the A-sentence itself stand in the requisite temporal relation
to each other. An A-sentence that locates an event in the present is
true if and only if the utterance of the A-sentence and the event
occur at the same time as each other. An A-sentence that locates an
event in the future is true if and only if the event occurs after the
utterance of the A-sentence. An A-sentence that locates an event in
the past is true if and only if the event occurs before the utterance of
the A-sentence. Facts about the temporal relations that obtain
between events and utterances about them are sufficient to account
for the truth of every true A-sentence.
The B-theory thus treats time in a way that is similar to our
treatment of space. When I say that 'London is here' I am not
attributing to London the property of being here. What makes my
utterance true, if it is true, is that the utterance occurs in the same

5
There are two different B-theoretic accounts of the facts that make A-
sentences true: the date version and the token-reflexive version. I argue
elsewhere (Dyke, forthcoming) that the token-reflexive version is prefer-
able. For the sake of simplicity, I only discuss the token-reflexive account
here.
146
 McTaggart and the Truth about Time

place as London. Similarly, when I say that 'it is now Autumn', I am

not attributing to Autumn the property of being present. What
makes my utterance true, if it is true, is that the utterance occurs at
the same time as Autumn.
A-sentences appear to change their truth-value over time. The
sentence 'The train is now arriving' is true at some times and false
at other times. According to the A-theory of time, the fact that A-
sentences change their truth-value over time reflects the fact that
events and states of affairs are continually changing their location in
the A-series. The reason why the sentence 'The train is now arriv-
ing' is sometimes true and sometimes false, is because the fact that
the train is now arriving is only a present fact at some times, but not
at others. It is only when it is a present fact that the sentence is true.
All this is denied by the B-theory of time.
The B-theory invokes the distinction between sentence-types and
sentence-tokens. A sentence-type has a 'changing' truth-value if
and only if some of its tokens are true and others false. Two tokens
of the sentence-type 'The train is now arriving' might have differ-
ent truth-values, but the truth-values they have are fixed and
unchanging. The truth-value that any token of this type has
depends on when it is produced. So, the claim that A-sentences
change their truth-value over time is wrong. The fact of the matter
is that some tensed sentence-types have some true and some false
tokens. This gives the impression that the sentence-type itself is a
determinate object with a changing truth-value, but sentence-
tokens are the proper bearers of truth and falsity, and their truth-
values are fixed and unchanging.

7. An objection to the token-reflexive version of the B-theory

The B-theory provides a token-reflexive account of the truth con-

ditions of A-sentences. According to this account, a token of an A-
sentence is true if and only if the event the A-sentence is about
stands in the appropriate temporal relation to the token of the A-
sentence itself. For example:
For any token u of 'The train arrived 2 hours ago' u is true if and
only if the train's arrival is 2 hours earlier than u.
The token-reflexive version is so called because the token itself con-
stitutes part of its own truth conditions. It intuitively delivers the
right truth conditions for tokens of A-sentences, but it has been
criticized on the grounds that there are some circumstances where
147
Heather Dyke

it delivers the wrong truth conditions.6 For example, William Lane

Craig (1996) argues that that 'The New B-theory can give no coher-
ent account of the truth conditions of tensed sentences which are
not tokened.' (Craig (1996) 18) He asks what the truth conditions
are for a sentence like 'There are no sentence-tokens now.' Any
token of this A-sentence-type would be false, because if the token
existed, the time at which it existed would not be a time of which it
is true to say that there are no tokens. However, it also seems to be
the case that there are some times of which it is true to say that there
are no tokens then. The point of this objection is that the sentence-
type 'There are no sentence-tokens' can express something true
even though no true token of it can ever be produced.
The force of this criticism stems from the intuition that truth, or
what is true of the world, does not depend on what anybody hap-
pens to say. But the token-reflexive version seems to imply that
truth depends on true tokens being produced. This criticism has
been articulated in some depth by Quentin Smith (1993), so I shall
address his statement of it. He says:
'If a normal A-sentence is used on some occasion to express
something true, what the A-sentence expressed on that occasion
would have been true then even if it had not been expressed.'
(Smith (1993) 83)
Smith is appealing to an intuition that we have about the concept of
truth. The intuition is that the way the world is does not depend on
there being utterances expressing that the world is that way. Smith
thinks that the token-reflexive theory is committed to the denial of
this intuition because it gives truth conditions that can only be ful-
filled when sentence-tokens are produced. I shall argue that the
token-reflexive theory is not defeated by this objection.
Suppose an event occurs, and lasts for a certain amount of time.
A forest fire starts at tx and burns itself out by t2. During that peri-
od of time no one utters any sentence that expresses that the forest
is now burning. Because the forest actually burns during this peri-
od of time, our intuition is that if someone had uttered such a sen-
tence it would have been true. But how can the token-reflexive
theory cohere with this intuition? I would explain it by putting for-
ward the following counterfactual: between t\ and t2, if someone
had uttered a token of the sentence-type 'The forest is now burn-
ing' , that token would have been true. The reason why it would have
been true is that its tenseless token-reflexive truth conditions would
6
See, for example, Smith (1993), Craig (1996) and Mellor (1998).
148
 McTaggart and the Truth about Time

have been satisfied. The truth conditions for a token of this sen-
tence-type are:
Any token u of 'The forest is now burning' is true if and only if
the burning of the forest is simultaneous with u.
In order for these truth conditions to be satisfied two events must
occur simultaneously: the burning of the forest and the production
of a token of the sentence-type. Between t\ and t2 the forest burns
so if, during that period of time, a token of the sentence-type is pro-
duced, its truth conditions would ipso facto be satisfied. However, if
no such token is produced, the forest still burns during that period
of time, but there is no token the truth or falsity of which we have
to account for.
The project of providing truth conditions for A-sentence-tokens
has both semantic and ontological significance.7 On the one hand, it
specifies what the world must be like in order for those tokens to be
true. This is its ontological function. If the truth conditions of A-
sentences only require the existence of B-facts, then that shows that
A-facts are not needed to account for the truth of A-sentence-
tokens. It also explicates how the truth or falsity of a sentence-token
depends^on what its semantic constituents mean when produced in
a given_gpntext. This is its semantic function. If the project is suc-
cessful it will show that the world need not be an A-world to
account for the fact that we sometimes utter true and meaningful A-
sentence-tokens, and it will also explain why the true A-sentence-
tokens we utter are true. The provision of truth conditions makes
perspicuous both the relationship between truth and reality, and
that between truth and meaning.
The concept of truth is connected both to meaning and to reality.
We might even say that it is ambiguous in that it has two distinct
domains of application. Linguistic entities are capable of being true
or false, and the world is that which makes true or false our utter-
ances about it. Consider the difference between the predicate 'true'
and the operator 'It is true that'.8 The predicate 'true' applies to lin-
guistic entities. It is sentence-tokens that can correctly be described
as true or false. However, if we prefix a sentence with 'It is true that'
we are making a claim about the world, not about the sentence. I can
describe the sentence 'The forest is now burning' as true or false.
But if I say 'It is true that the forest is now burning' I am making a
7
Davidson (1986) recognizes both kinds of significance when he says
'The truth of an utterance depends on just two things: what the words as
spoken mean, and how the world is arranged.' (Davidson (1986) 309)
8
I am grateful to Colin Cheyne for suggesting this explanation to me.
149
Heather Dyke

claim about what the world is like; I am describing reality, not a sen-
tence about it.
It is important to be clear, when expressing one's intuitions about
truth, whether those intuitions are about the connection between
truth and meaning or that between truth and reality. It is the con-
nection between truth and reality that generates our intuition that
truth is independent of the production of any sentence-tokens. The
world is the way it is independently of what anyone happens to say
about it. This is the intuition that Smith thinks the token-reflexive
theory is unable to explain. His example constitutes a sentence-
type, a token of which would have been true if it had been uttered
at a certain time, but no such token was uttered. He argues that the
token-reflexive theory cannot account for our intuition that this
sentence expresses a truth whether or not a token of it is produced.
But the token-reflexive theory can account for this intuition, simply
by upholding the distinction between the ontological and the
semantic aspects of truth. We have an intuition that the sentence
'The forest is now burning' is true between tx and t2. The intuition
can be explained by appealing just to the ontological aspect of truth.
Between t^ and t2 it is true that the forest is burning, but if no sen-
tence-token is produced, there is nothing to which we can ascribe
the predicate 'true'. But reality remains the same whether or not
sentences about it are produced.
Lastly, to return to the original challenge, I must explain how I
would assign truth conditions to the sentence-type 'There are no
sentence-tokens now' of which there can be no true tokens, even
though what it expresses can be true. The general truth-conditional
formula for this sentence-type is:
Any token, u, of 'There are no sentence-tokens now' is true if and
only if u occurs at a time at which there are no sentence-tokens.
Now, consider some arbitrary token of that sentence-type. The time
at which it is produced cannot be a time at which there are no sen-
tence-tokens, so no true token of that sentence-type can be uttered.
And the token-reflexive analysis explains why this is the case.
However, if we turn now to the ontological aspect of truth, we can
see that reality can be such that there are times at which it is devoid
of sentence-tokens. That is all that is meant by the claim that there
are times at which this sentence, or what it expresses, is true. It is,
indeed, misleading to say that there are times at which this sentence
is true. It is misleading in two ways. Firstly, it is ambiguous between
sentence-types and sentence-tokens. As I have argued, it is sen-
tence-tokens, not sentence-types that have truth-values, and there
150
 McTaggart and the Truth about Time

can be no times at which a token of this sentence can be true.

Secondly, it equivocates between the semantic and the ontological
aspects of truth. There are times at which it is true that there are no
sentence-tokens, but there are no tokens of this sentence-type that
can be described as true.
I have only dealt with one objection to the token-reflexive version
of the B-theory of time, but it is, I believe, one of the most com-
pelling. Those who reject this theory, very often do so on the
grounds that it cannot account for the truth of unuttered proposi-
tions.9 By restricting the application of the predicate 'true' to sen-
tence-tokens, and by upholding the distinction between the seman-
tic and the ontological aspects of truth, the problem for the token-
reflexive theory evaporates. Furthermore, I hope to have deflected
the criticism that the token-reflexive theory has the unacceptable
consequence that the way the world is depends on what human
beings happen to say about it.

8. Conclusion

My hope in presenting this paper has been to guide you all on a

journey, starting from McTaggart's argument for the unreality of
time, and finishing up with what I believe to be the truth about
time. I think McTaggart was right to argue that our world cannot
be an A-world: it cannot be a world in which anything is really past,
present or future. But I think he was wrong to argue that our world
has to be an A-world if time itself is to be a part of it. Our world is
a B-world, in spite of the fact that we talk and think as if it were an
A-world. Indeed, the fact that we live in a B-world can provide the
best explanation for the truth of our true A-sentences.

References

Bigelow, J. 1996. 'Presentism and Properties', in James E. Tomberlin (ed.)

Philosophical Perspectives 10, Metaphysics, (Oxford: Blackwell
Publishers Ltd, 1996) 35-52.
Craig, W. L. 1996. 'Tense and the New B-theory of Language', Philosophy
71, 5-26.
1998. 'McTaggart's Paradox and the Problem of Temporary
Intrinsics', Analysis 58, 122-27.
9
Mellor (1998), for example, rejects the token-reflexive theory in favour
of the date theory for just this reason.
151
Heather Dyke

Davidson, D. 'A Coherence Theory of Truth and Knowledge', in Ernest

LePore (ed.) Truth and Interpretation: Perspectives on the Philosophy of
Donald Davidson (Oxford: Basil Blackwell, 1986) 307-19.
Dummett, M. 1960. 'A Defense of McTaggart's Proof of the Unreality of
Time', Philosophical Review 69, 497-504.
Dyke, H. (forthcoming, 2002) 'Tokens, Dates and Tenseless Truth
Conditions', Synthese.
Hinchliff, M. 1996. 'The Puzzle of Change' in James E. Tomberlin (ed.)
Philosophical Perspectives 10, Metaphysics, (Oxford: Blackwell
Publishers Ltd, 1996) 119-36.
McTaggart, J. M. E. The Nature of Existence, Vol II, (Cambridge:
Cambridge University Press, 1927)
Mellor, D. H. Real Time II, (London: Routledge, 1998)
Prior, A. N. 1970. 'The Notion of the Present', Studium Generale 23,
245-48.
Smith, Q. Language and Time, (New York: Oxford University Press, 1993)
Sprigge, T. L. S. 1992. 'The Unreality of Time', Proceedings of the
Aristotelian Society 92, 1-19.

152
On Absolute Becoming and the Myth
of Passage
STEVEN F. SAVITT

In the literature on time in the twentieth century stemming from J.

M. E. McTaggart's famous argument for the unreality of time,1 two
gems stand out. The first is C. D. Broad's patient dissection of
McTaggart's argument in the chapter 'Ostensible Temporality' in
his Examination of McTaggart's Philosophy.1 Broad carefully, and to
my mind persuasively, uncovers the root errors in McTaggart's
argument. In addition he tentatively proposes that the features of
time that he calls its transitory aspect can be explained in terms of
a dynamic aspect of time that he calls Absolute Becoming.
The second gem is D. C. Williams' paper, 'The Myth of Passage,'
a gloriously over-written rant against the idea that there is some-
thing active or dynamic to time over and above 'the spread of events
in space-time'. 3 Broad is mentioned thrice as a proponent of this
myth. His contrast of the transitory aspect of time to its extensive
aspect and his invocation of Absolute Becoming are mentioned in
Williams' survey of attempts to characterize passage. (102-4) A few
paragraphs later Broad is specifically mentioned, along with
Bergson and Whitehead, as trying but failing to escape 'the para-
doxes of passage'. (106)
Broad clearly is on Williams' enemies list, and the general intent,
as well as style, of Broad's chapter and Williams' paper could
1
The argument first appeared in J. M. E. McTaggart's 'The Unreality
of Time' Mind, New Series, No. 68 (October, 1908). A later version of this
argument appears as Chapter 33 in McTaggart's The Nature of Existence,
Vol. II (Cambridge University Press, 1927).
2
'Ostensible Temporality' is chapter 35 of Volume II of Broad's
Examination of McTaggart's Philosophy, first published by Cambridge
University Press in 1938 and reprinted, with the same pagination, by
Octagon Books in 1976. References to Broad in the text will, unless other-
wise specified, be to 'Ostensible Temporality'.
3
Williams' paper first appeared in Journal of Philosophy 48 (1951). Page
references in the text to this paper will be to the reprint in Richard Gale's
The Philosophy of Time (Anchor Books, Doubleday and Company, Inc.,
1967). The quote above appears on p. 99.
153
Steven F. Savitt

scarcely be more opposed. There is, nevertheless, an area of con-

vergence, or even overlap, between the views of Broad and Williams
that has not been remarked and that may help clarify the nature of
passage. My aim in this paper is to indicate the nature and impor-
tance of this surprising common ground.

I. True and Literal Passage

In 'The Myth of Passage' Williams commends his view, 'the

doctrine of the manifold', as an antidote to passage views, but he
does little to describe this 'doctrine'. His most explicit statement of
it comes in a later exchange with Milic Capek.
What I advocate as 'the doctrine of the manifold... is simply a
philosophical acceptance, as an ultimate literal truth about the
way things are in themselves, of the conception that nature, all
there is, was, or will be, 'is' (tenselessly) spread out in a four-
dimensional scheme of location relations which intrinsically are
exactly the same, and hence in principle commensurate, in all
directions, but which happen to be differentiated, in our neigh-
bourhood at least, by the de facto pattern of the things and events
in them—by the lie of the land, so to speak. We are all perfectly
familiar with the fact that the prodigious difference of the verti-
cal dimension of space, with its terrifying asymmetry of up and
down, above and below, from all those comparatively indifferent
directions we call horizontal, is not due to any intrinsic difference
between vertical and horizontal distances but only to a certain
characteristic complex of matter and force in our vicinity whose
'grain,' so to speak, runs one way and not the other. Just so, I
argue, there is a somewhat more pervasive pattern of physical
qualities and relations which constitutes the even more momen-
tous oddity of the temporal direction, with its even more striking
asymmetry of earlier and later, in contrast with all the so-called
spatial directions. Very much as the singularities of arrangement
which distinguish the vertical from the horizontal were explained
by Descartes and Newton, so recent scientific attention to 'the
direction of time' has begun the description of the arrangements
which distinguish the temporal from the merely spatial.4

4
Donald Williams, 'Physics and Flux: Comment on Professor Capek's
Essay' in Boston Studies in the Philosophy of Science, Volume II
(Humanities Press, 1965): 465-66.
154
 On Absolute Becoming and the Myth of Passage

It is a nice project, which I will engage in only superficially, to make

sense of this doctrine. At some points Williams seems to take the equiv-
alence of the four dimensions quite literally. He writes, for instance:
It is conceivable, then, though perhaps physically impossible, that
one four-dimensional part of the manifold of events be slued
around at right angles to the rest, so that the time order of that
area, as composed of its interior lines of strain and structure, runs
parallel with a spatial order in its environment. It is conceivable,
indeed, that a whole human life should lie thwartwise of the man-
ifold, with its belly plump in time, its birth in the east and its
death in the west, and its conscious stream perhaps running
alongside somebody's garden path. (112)5
At other places he makes important claims that seem to me incon-
sistent with the above. For instance:
The term 'the present' is the conventional way of designating the
cross section of events which are simultaneous with the uttering of
the phrase, and 'the present moves' only in that when similar words
occur at successively different moments, they denote, by a twist of
language essentially the same as that of all 'egocentric particulars'
like 'here' and 'this', different cross sections of the manifold. (105)
In my view Williams would not be able to use the definite descrip-
tion 'the cross section of events' at some time t if there were not one
distinguished (temporal) dimension of the manifold from which to
locate a unique corresponding three-dimensional set of simultane-
ous events. So, while noting that Williams' official doctrine of the
manifold runs at some points counter to it, I am going to write
below as if Williams shared in a view about the structure of space-
time held more or less articulately by almost all the participants in
the debate concerning passage stemming from McTaggart. In this
view the basic elements of the spacetime ontology are instantaneous
events.6 These events may be sorted into equivalence classes of
5
Consider the following remark of Julian Barbour's on page 242 of The
End of Time: The Next Revolution in Physics (Oxford, 1999): 'The coordi-
nates laid down on space-time are arbitrary. Since the coordinates include
one used to label space-time in the time direction and all coordinates can
be changed at whim, there is clearly no distinguished label of time.'
6
Some use the term event to refer also to sets of events extended in space
and time, like World War II. I prefer to use the term process for such sets
of events. The events considered here have or occur at spatiotemporal
locations and may have causal relations to one another. In spacetime
theories, the spatiotemporal locations themselves are called events as well.
155
Steven F. Savitt

simultaneous events and these classes (which I will call moments or

times, without wishing to prejudge any questions in the substanti-
valism/relationism controversy) can be completely ordered by the
asymmetric and transitive binary relation is earlier than or its con-
verse is later than.
The defining feature of events is that they happen. As Williams
puts it:
'Taking place' is not a formality to which an event incidentally
submits—it is the event's very being. World history consists of
actual concrete happenings in a temporal sequence... (106)
Williams is a naturalist. The manifold reflects all and only what
there (basically) is. It's hard to resist quoting his own way of putting
it:
I believe that the universe consists, without residue, of the spread
of events in space-time, and that if we thus accept realistically the
four-dimensional fabric of juxtaposed actualities we can dispense
with all those dim nonfactual categories which have so bedeviled
our race: the potential, the subsistential, and the influential, the
noumenal, the numinous, and the nonnatural. But I am arguing
here, not that there is nothing outside the natural world of events,
but that the theory of the manifold is anyhow literally true and
adequate to that world; true, in that the world contains no less
than the manifold; adequate, in that it contains no more. (99)
Many (and Williams marshals an impressive list) have thought that
the manifold missed an important extra ingredient of our world, the
passage of time. 'This something extra...' writes Williams, 'is a
myth: not one of those myths which foreshadow a difficult truth in
a metaphorical way, but altogether a false start, deceiving us about
the facts, and blocking our understanding of them.' (102) After
burying the reader in a near avalanche of evocative quotations try-
ing to express the idea of passage,7 Williams comments that 'the
instant one thinks about them one feels uneasy, and the most labo-
rious effort cannot construct an intelligible theory which admits of
the literal truth of any of them.' (104) Passage is, to repeat, 'an
altogether false start'.
There is, however, a thread that runs through Williams' essay that
has not drawn much comment. Consider the following remarks

7
My favourite is from Santayana: 'The essence of nowness runs like fire
along the fuse of time.'
156
 On Absolute Becoming and the Myth of Passage

• [T]he theory of the manifold provides the true and literal

description of what the enthusiastic metaphors of passage
have deceptively garbled. (109)
• [T]he dimensional theory accommodates what is true in the
notion of passage, that is, the occurrence of events, in con-
trast with a mythical rearing and charging of time itself...
(113)
• There is passage, but it is nothing extra. It is the mere
happening of things, their existence strung along in the
manifold. (105)

It is not too difficult to see what Williams is getting at. Events in

the manifold occur at times, some simultaneously, some earlier
than others, some later. True and literal passage is the ordered
occurrence of (simultaneity sets of) events in the manifold. If the
manifold does have a temporal dimension (as I have argued it does,
despite the occasional remark of Williams to the contrary), then it
can accommodate or model or provide a representation of this true
and literal idea of passage. Indeed, it can hardly avoid so doing. Is
this true and literal passage, however, truly and literally passage—
the real whooshy, zingy thing that is so salient in our experience?
Keeping that question in mind, let us turn to the enemies list.

II. The Transitory Aspect of Temporal Facts

Whereas Williams barely mentions McTaggart's argument for the

unreality of time, C. D. Broad's chapter 'Ostensible Temporality' is
a full-dress exegesis and criticism of that argument.8 McTaggart's
argument is, in essence, this: there must be passage if there is to be
change and there must be change if there is to be time; but passage
is a self-contradictory notion, and hence there is no time.
Broad deals with this argument in great (excruciating, some
would say) detail. In all this detail, it may be that the main thread
of Broad's argument is occasionally lost. McTaggart, according to
Broad, does present a way of construing passage—as events con-
stantly changing with respect to properties like presentness, past-
ness, and futurity—that is indeed incoherent, but Broad argues that

8
As only befits a chapter, Chapter 35 of Volume II, of a massive work
entitled Examination of McTaggart's Philosophy.
157
Steven F. Savitt

there is a distinct notion of passage, Absolute Becoming, that is

not.9
The route to Absolute Becoming starts in a distinction between
those aspects of time in which it is like space (duration being like
extension), which Broad calls the extensive aspect of temporal facts,
and a peculiar aspect of time in which it seems very different from
space, the transitory aspect of temporal facts.
The third, and much the most puzzling, set of temporal charac-
teristics are those which are involved in facts of the following
kind. An experience is at one time wholly in the future, as when
one says 'I am going to have a painful experience at the dentist's
tomorrow' It keeps on becoming less and less remotely future.
Eventually the earliest phase of it becomes present; as when the
dentist begins drilling one's tooth, and one thinks or says 'The
painful experience I have been anticipating has now begun.' Each
phase ceases to be present, slips into the immediate past, and then
keeps on becoming more and more remotely past. But it is fol-
lowed by phases which were future and have become present.
Eventually the latest phase of this particular experience becomes
present and then slips into the immediate past. There is the fact
which one records by saying 'Thank God (on the theistic hypoth-
esis) that's over now!' After that the experience as a whole retreats
continually into the more and more remote past. [266-67]

McTaggart takes these facts at face value and treats passage as a

kind of qualitative change—events changing with respect to prop-
erties like presentness, the various degrees of pastness, and the var-
ious degrees of futurity. Broad points out, however, that qualitative
change requires the existence of some thing—in this case, an
event—that changes its properties by persisting through time. Since
events are by definition instantaneous, it is awkward to suppose also
that they persist through time. In order to avoid this awkwardness,

9
I find the essentials of this argument in section 1.22, 'Absolute
Becoming,' of Broad's chapter. That section appears in part I,
'Independent Account of the Phenomenology of Time,' before
McTaggart's argument officially enters the stage in Parts II and III, and
hence it is easy to see how its importance might be overlooked. I ask those
who doubt my reading of Broad to reflect on Broad's remark, which seems
to appear out of the blue at the very end of his consideration of
McTaggart's main argument, that '[t]he fallacy in McTaggart's argument
consists in treating absolute becoming as if it were a species of qualitative
change...' (317)
158
 On Absolute Becoming and the Myth of Passage

a defender of the idea that passage is a kind of qualitative change is

inevitably tempted to suppose that events do persist (so they can
change their properties) but that this persistence is not in ordinary
time (in which they are instantaneous). The persistence must then
be in a second temporal dimension: but if this second dimension
truly is temporal, it too must admit of passage and the construction
that boosted us from the first to the second temporal dimension
bids fair to push us on to a third. Infinite regress beckons.
Broad's response was to propose that we not take the transitory
aspect of temporal facts at quite face value:
When one finds oneself launched on an endless series of this kind
it is generally a sign that one has made a false move at the begin-
ning. I think it is easy to see what the false move is in this case.
The phrase 'to become present' is grammatically of the same
form as the phrase 'to become hot' or 'to become louder'. We are
therefore tempted to think that sentences like 'This event became
present' record facts of the same kind as those which are record-
ed by sentences like 'This water became hot' or 'This noise
became louder.' Now a very little reflection is enough to show
that this is a mistake. (280)
The mistake is to treat passage as like qualitative change.10 What
kind of fact, then, is recorded by sentences like 'E became present'
if not the acquisition by an event E of a new property? Here is
Broad's answer:
But a literally instantaneous event-particle can significantly be
said to 'become present'; and, indeed, in the strict sense of 'pre-
sent' only instantaneous event-particles can be said to 'become
present'. To 'become present' is, in fact, just to 'become', in an
absolute sense; i.e., to 'come to pass' in the Biblical phraseology,
or, most simply, to 'happen'. Sentences like 'This water became
hot' or 'This noise became louder' record facts of qualitative
change. Sentences like 'This event became present' record facts
of 'absolute becoming'.

Absolute becoming, as explained by Broad, is just the happening of

events. Since events are located at various times or moments, they
happen at various times or moments. Some events have happened,
some are happening now, and others, we hope, will happen eventu-
ally. Some events occur simultaneously, some earlier than others,
10
There is a parallel argument in the text that one should not treat pas-
sage as like motion either.
159
Steven F. Savitt

some later. Absolute becoming is the ordered occurrence of (simul-

taneity sets of) events. This is how Broad proposed that we should
think of passage, and, as far as I can see, there is no difference what-
soever between his understanding of absolute becoming and Williams'
true and literal becoming. Generally speaking, I think it is more use-
ful to distinguish than to assimilate positions. In this particular case,
however, the existence of a surprising common ground shared by a
resolutely anti-passage philosopher like Williams and a stalwart (at
least at this period of his life) pro-passage philosopher like Broad
must be a significant clue to understanding the nature of temporal
becoming.
Since Williams objected vociferously to Broad's absolute
becoming, might it not be thought preposterous to propose that it
is none other than his own true and literal becoming? Williams
objects (106) that 'the extra idea of passage or absolute becoming'
leads one to postulate an infinite hierarchy of times. Then
Williams complains that absolute becoming 'involves the same
anomalies of metahappening and metatime which we observed in
the other version.' (107) Which other version Williams intends is
not entirely clear, but if it is a version which treats passage as a
sort of motion or as a kind of qualitative change, the argument is
beside the point. Broad, as noted, agrees that treating passage as
qualitative change or motion leads to contradiction or regress, and
he proposes a different approach. Of course it is possible that
Broad's own way of understanding passage leads to contradiction
or regress too, but Williams does not address this possibility, rely-
ing on the old arguments against the old views. T h e burden of
proof is on Williams at this point, but he does not shoulder this
burden at all.
For a second problem with my irenic thesis that is bound to occur
to many, note that Williams wrote, 'The statement that a sea fight
not present in time nevertheless exists is no more contradictory than
that one not present in space nevertheless exists.' (101) One would
suppose that Broad differed here, that he would have denied the
existence of tomorrow's sea fight." Is that not an indication that
11
Broad clearly did so in his treatment of time in Scientific Thought, in
which he defended the idea that the future is nothing but that once an
event becomes (or happens) then it continues to exist forever. I take Broad
to have dropped the latter half of this view of absolute becoming by the
time he wrote Examination of McTaggart's Philosophy, given his explicitly
deflationary line that for an event to become absolutely is just for it to
happen.
I find support for my claim, not just in the text of Examination of
160
 On Absolute Becoming and the Myth of Passage

Broad 'took passage seriously' whereas Williams did not? How,

then, could they have the same concept of passage?
I suggest that the alleged difference expressed in the previous
paragraph is no difference at all, but merely a verbal confusion.
There is an ordinary tensed sense of 'is' or 'exists' which, in the case
of events, simply indicates that they are happening or occurring or
that their appointed moment has arrived. Whatever time t it is now,
both Broad and Williams agree that all and only those events that
occur at t exist (that is, occur). Williams ought not be saddled with
believing that tomorrow's sea fight occurs today.
In thinking of the sea fight, Williams probably slipped into using
a tenseless sense of 'exist' in which an event EXISTS (as I will
write to indicate the tenseless sense) if it has happened, is happen-
ing, or will happen. In this sense, tomorrow's sea fight does EXIST,
and Broad should not be saddled with the view that it does not
EXIST—that is, is not going to happen. If there is a difference
between Broad and Williams about the existence of the past and
future, then it lies much more deeply than this objection contem-
plates. If it lies deep, it may well be entirely unconnected to their
shared view of passage.

III. Defending the Radical Middle

I propose that Williams' true and literal passage and Broad's

Absolute Becoming is all there is to the passage of time. While I

McTaggart's Philosophy but also in Broad's later statement (on pp. 766-7
of 'A Reply to My Critics' in The Philosophy of C. D. Broad, edited by P.
A. Schilpp (Tudor Publishing Company, 1959)) that though he once took
seriously the idea of the world's history 'growing continually longer in
duration by the addition of new slices,' he now lumped that idea in with
the 'policeman's bulls-eye' metaphor as an inadequate way of trying to
understand absolute becoming.
I have no idea, by the way, whether Williams ever noticed this passage and
commented on it. Any information on this point would be greatly appreciat-
ed. In his earlier discussion of Broad's views on time in 'The Sea Fight
Tomorrow' (in Structure, Method and Meaning: Essays in Honor of Henry
M. Sheffer, edited by Paul Henle, Horace M. Kallen, and Suzanne Langer
(The Liberal Arts Press, 1951)) Williams noted that Broad had indicated
that his views in Examination of McTaggart's Philosophy differed in some
ways from those of Scientific Thought. Williams quite understandably took
it to be probable that the absolute becoming of the later work should be
understood as the becoming of the earlier work, and I suspect that many
readers have (mistakenly, in my view) read Broad through his eyes.
161
Steven F. Savitt

hope that my proposal will satisfy both sides in the dispute over the
existence of passage, I fear it will be acceptable to neither. Where,
the proponents of passage will ask, is the whiz and go in true and
literal passage? How, the opponents of passage will wonder, can
absolute becoming not engender paradox or inconsistency?
I have supposed that this investigation of the nature of passage
takes place within a given presupposed background spacetime
structure—that events can be partitioned into equivalence classes of
simultaneous events that then can be completely ordered by an
asymmetric and transitive relation such as is earlier than. The
unease with my suggestion that true and literal passage is all there
is to passage is that it comes close to merely pointing to this space-
time structure, which, it is often said, is a changeless structure and
so hopeless as a model for change or passage. It is always true to say,
it will be objected, that the events that occur at t0 occur at t0, always
true to say that the events that occur at £j occur at tit and always true
to say that the events that occur at t0 occur earlier than the events
that occur at tlt and so on. These unchanging facts are often illus-
trated by a picture like the one below, with two-dimensional planes
replacing three-dimensional spaces for ease of illustration. Where in
this picture, it may be asked, is passage?
My response to this critical question is that change is not in this
picture but in what it is a picture of. One who asks it is confusing a

Successive sets of simultaneous events

162
 On Absolute Becoming and the Myth of Passage

static representation with a representation of stasis. If we learn

from philosophers of mind that 'we must distinguish features of
representmgs from the features of represented', 12 if we keep in mind
that one dimension of this spacetime structure is supposed to rep-
resent time, that events occur at times, and that different events
occur at different times, I think we should have no trouble in under-
standing that this static structure can represent a dynamic or
unfolding world.'3 We do not need an animated picture to have a
picture of animation. In my view, the call for animation in the
model (rather than in what it is a model of) is an unnecessary dupli-
cation that is at the root of most of the paradoxes and regresses that
are supposed to attend upon the idea of passage.14
In a very important paper, 'The Meaning of Time,' 15 Adolf
Griinbaum argued that becoming or passage is mind-dependent.
But he explicitly contrasted passage with events happening seri-
atim. He wrote:
It is... this occurring now or coming into being of previously
future events and their subsequent belonging to the past which is
called 'becoming' or 'passage'. Thus, by involving reference to
present occurrence, becoming involves more than mere occur-
rence at various serially ordered clock times.16
Griinbaum adds that 'nowness, in the sense associated with becom-
ing, plays no role as a property of physical events in any of the
extant theories of physics.'17 In the picture on the previous page
12
Consciousness Explained by Daniel Dennett (Little, Brown and
Company, 1991), p. 147.
13
Palle Yourgrau (on p. 22 of Godel meets Einstein (Open Court, 1999))
says that 'for Godel, to spatialize time is to render it ideal (by robbing it of
its characteristic mode of existence).' I think that Yourgau concurs in
Godel's view, and these two are but part of a chorus who have complained
that spatializing time ignores its essential characteristic, passage. If what I
say in the text is correct, however, one can do full justice to both the exten-
sive aspect of time (that is, spatialize it) and to its transitory aspect, at least
in the setting of classical spacetime.
14
1 am here consciously echoing the language of David Park in his essay
'The Myth of the Passage of Time' in The Study of Time: Proceedings of
the First Conference of the International Society for the Study of Time,
edited by J. T. Fraser, F. C. Haber, and G. H. Muller (Springer-Verlag,
1972). This paper originally appeared in Studium Generale 24 (1971):
19-30.
15
In Basic Issues in the Philosophy of Time, Eugene Freeman and Wilfrid
Sellars (eds.) (Open Court, 1971), pp. 195-228.
16
'The Meaning of Time', p. 195.
"Ibid, p. 211.
163
Steven F. Savitt

derived from classical physics, there is indeed no noweness to be

seen. But, as I have claimed, it is all too easy to miss the
becoming.
Grunbaum and I agree, I think, on what is the case. We agree that
events occur at various serially ordered clock times, and we also
agree that nowness is not a property that hops from event to event.
What we disagree on seems close to emphasis, for I would drop the
'mere' in his second sentence and would then explain the absence of
nowness from the mind-independent world by treating 'now' as an
indexical, like 'here'. 18
With respect to objections from the opponents of passage, who
believe that the idea of passage involves paradox or regress, I have
tipped my hand above. I think all such objections are objections to
models of passage that construe it as a kind of motion or as qualita-
tive change. Such objections, I believe, do not apply to the modest
conception of passage as the successive happening of events advo-
cated above.
Passage deniers might fret that the account of passage above is
too modest—that it is, in fact, threadbare because it assumes too
much or leaves too much unexplained. I identify passage with
absolute becoming and Broad wrote, 'I do not suppose that so sim-
ple and fundamental a notion as that of absolute becoming can be
analysed...' (281) As I understand Broad, he is saying that the hap-
pening of events is so fundamental a notion that it cannot be
explained in terms of simpler or more basic ideas, and I have con-
siderable sympathy with this claim. The authors of the entry on
Time in The Cambridge Dictionary of Philosophy note, with disap-
proval I think, 'Broad attempted to skirt these perplexities [in giv-
ing an account of passage] by saying that becoming is sui generis
and thereby defies analysis, which puts him on the side of the mys-
tically inclined Bergson...' I've not noted a mystical streak in
Broad, and I do think that one has to start somewhere, has to have
some primitives. If and when I can say more about the happening
of events, I will; but until then it does not seem to me either
unreasonable or mystical or question-begging to start with it as a
basic notion.19
18
Of course, indexical expressions like 'now' can be used only in cir-
cumstances much like those that Grunbaum took to show that becoming is
mind-dependent—especially if there is a close connection between con-
ceptualization and possession of a language.
19
Incidentally, if I can infer from the above remark that I have both
Broad and Bergson on side for my account of passage, what more could I
need to establish its bona fides to supporters of passage?
164
 On Absolute Becoming and the Myth of Passage

Passage affirmers may also find my account of passage too mod-

est in another way. I suspect that J. J. C. Smart was not exaggerat-
ing when he wrote, at the conclusion of a chapter purporting to
show that there was no genuine passage or becoming:
In this chapter I have been defending the view of the world as a
four-dimensional system of entities in space-time. Concepts such
as 'past' and 'future' have been shown to be anthropocentric in
that they relate to particular human utterances. My advocacy of
the four-dimensional picture of the world is, therefore, among
other things, part of the same campaign against anthropocentric-
ity and romanticism in metaphysics that I have been waging else-
where... It is surely no accident that romantic, vitalistic, and anti-
mechanistic philosophies such as those of Bergson and
Whitehead are also those which lay great emphasis on the alleged
transitory aspect of time, process or absolute becoming. While I
concede that our present notions of space and time may perhaps
have to be revised, the idea of the world as a space-time manifold
is nearer the truth than these romantic and obscure philosophical
theories.20
When fancy verbiage, like 'Absolute Becoming', is stripped away,
the notion of temporal becoming defended here is rather plain,
homespun, humdrum, deflated, dowdy. There is no special connec-
tion between this sort of passage and either freedom, spontaneity,
and emergence on the one hand or determinism, necessity, and
reductionism on the other. It neither supports nor discourages
romanticism and may therefore disappoint those who feel that pas-
sage must be portentous. It is root and branch neutral but, in my
view, none the worse for that.

IV. Conclusions
I have argued that there is a kind of passage or temporal becoming
espoused by at least one classic proponent of passage and admitted
by at least one classic opponent of passage. This common or garden
variety of passage is, I suggest, robust enough to satisfy those who
insist on the dynamic or transitory aspect of time, yet weak enough
to avoid any metaphysical or a priori objections. If this suggestion
is correct, then one no longer need try to construe passage in terms

20
Smart, J. J. C, Philosophy and Scientific Realism (New York: The
Humanities Press, 1963), p. 148.
165
Steven F. Savitt

of qualitative change or motion.21 If this suggestion is incorrect,

then one is left with either the task of reducing this minimalist
notion of passage even further to make it acceptable or, more likely,
expanding it in some way that yet avoids the traditional metaphysi-
cal pitfalls.
Recognizing passage as no more (or less) than absolute becoming
may help to solve some problems. I have suggested above, for
instance, that it provides a helpful way for looking at questions
about the reality of past and future (though I have by no means
tried to provide a complete discussion of the issues involved). Also,
in a recent paper,22 Clifford Williams claims that Bergsonian intu-
ition can not distinguish between classical passage and non-passage
metaphysics. If these two ostensibly opposed metaphysics are each
committed to a minimal common concept of transience, then it is no
mystery why Bergsonian intuition should fail to find a difference.
Good solutions raise problems too. If it does turn out that
absolute becoming is the best way to understand passage in the
spacetime structure described at the beginning section III of this
paper and presupposed throughout, then passage becomes mysteri-
ous again as soon as one turns to the spacetime of special relativity
and, perhaps, to other more general spacetimes. Minkowski space-
time could scarcely be less hospitable to absolute becoming, since its
geometry does not admit a unique partitioning into the sets of
simultaneous events needed to occur successively.23 General
relativistic spacetimes most likely do not admit any privileged
partitioning either. Can a differentiable manifold, then, provide 'the
true and literal description of what the enthusiastic metaphors of
passage have deceptively garbled', as Williams claimed his 'theory
of the manifold' could?
21
In any case, Broad's arguments against construing passage in these
ways seem quite difficult to evade. McTaggart's own regress argument,
however, may not be so formidable. I have tried to show why it fails in 'A
Limited Defense of Passage', American Philosophical Quarterly 38 (2001),
261-70,.
22
'A Bergsonian Approach to A- and B-Time,' Philosophy 73 (1998):
379-93. See also his 'The Metaphysics of A- and B-Time,' The
Philosophical Quarterly (1996): 371-81. I am in considerable sympathy
with the ideas expressed in these papers, though I do not wish to express
them in terms of Bergsonian intuition or in terms of opposition between
A-theories and B-theories of time.
23
I discuss the difficulties of importing the metaphysics of presentism
into Minkowski spacetime in 'There's no time like the present (in
Minkowski spacetime),' Philosophy of Science 67 (2000; Proceedings):
S663-S574.
166
 On Absolute Becoming and the Myth of Passage

This question seems to be the important and puzzling question

concerning passage. I hope this paper helps to make it more central
to philosophy of time in the 21 st century than it was in the 20th.

167
Time Travel and Modern Physics
FRANK ARNTZENIUS AND TIM MAUDLIN

Time travel has been a staple of science fiction. With the advent of
general relativity it has been entertained by serious physicists. But,
especially in the philosophy literature, there have been arguments
that time travel is inherently paradoxical. The most famous paradox
is the grandfather paradox: you travel back in time and kill your
grandfather, thereby preventing your own existence. To avoid
inconsistency some circumstance will have to occur which makes
you fail in this attempt to kill your grandfather. Doesn't this require
some implausible constraint on otherwise unrelated circumstances?
We examine such worries in the context of modern physics.

A Botched Suicide
You are very depressed. You are suicidally depressed. You have a
gun. But you do not quite have the courage to point the gun at your-
self and kill yourself in this way. If only someone else would kill
you, that would be a good thing. But you can't really ask someone to
kill you. That wouldn't be fair. You decide that if you remain this
depressed and you find a time machine, you will travel back in time
to just about now, and kill your earlier self. That would be good. In
that way you even would get rid of the depressing time you will
spend between now and when you would get into that time machine.
You start to muse about the coherence of this idea, when something
amazing happens. Out of nowhere you suddenly see someone com-
ing towards you with a gun pointed at you. In fact he looks very
much like you, except that he is bleeding badly from his left eye, and
can barely stand up straight. You are at peace. You look straight at
him, calmly. He shoots. You feel a searing pain in your left eye. Your
mind is in chaos, you stagger around and accidentally enter a
strange looking cubicle. You drift off into unconsciousness. After a
while, you can not tell how long, you drift back into consciousness
and stagger out of the cubicle. You see someone in the distance
looking at you calmly and fixedly. You realize that it is your younger
self. He looks straight at you. You are in terrible pain. You have to
end this, you have to kill him, really kill him once and for all. You
shoot him, but your eyesight is so bad that your aim is off. You do
169
Frank Arntzenius and Tim Maudlin

not kill him, you merely damage his left eye. He staggers off. You
fall to the ground in agony, and decide to study the paradoxes of
time travel more seriously.

Why Do Time Travel Suicides Get Botched?

The standard worry about time travel is that it allows one to go back
and kill one's younger self and thereby create paradox. More general-
ly it allows for people or objects to travel back in time and to cause
events in the past that are inconsistent with what in fact happened.
(See e.g. Godel 1949, Earman 1972, Malament 1985a&b, Horwich
1987). A stone-walling response to this worry is that by logic indeed
inconsistent events can not both happen. Thus in fact all such schemes
to create paradox are logically bound to fail. So what's the worry?
Well, one worry is the question as to why such schemes always
fail. Doesn't the necessity of such failures put prima facie unusual
and unexpected constraints on the actions of people, or objects, that
have travelled in time? Don't we have good reason to believe that
there are no such constraints (in our world) and thus that there is no
time travel (in our world)? We will later return to the issue of the
palatability of such constraints, but first we want to discuss an argu-
ment that no constraints are imposed by time travel.

Topology and Constraints

Wheeler and Feynman (1949) were the first to claim that the fact
that nature is continuous could be used to argue that causal influ-
ences from later events to earlier events, as are made possible by
time travel, will not lead to paradox without the need for any con-
straints. Maudlin (1990) showed how to make their argument pre-
cise and more general, and argued that nonetheless it was not com-
pletely general.
Imagine the following set-up. We have a camera ready to take a
black and white picture of whatever comes out of the time machine.
The film is then developed and the developed negative is subse-
quently put in the time machine and set to come out of the time
machine at the time the picture is taken. This surely will create a
paradox: the negative will have the opposite distribution of black,
white, and shades of grey, from the picture that comes out of the
time machine. But since the thing that comes out of the time
machine is the negative itself it we surely have a paradox.
170
 Time Travel and Modern Physics

However, it does not take much thought to realize that there is no

paradox here. What will happen is that a uniformly grey picture will
emerge which produces a negative that has exactly the same uni-
form shade of grey. No matter what the sensitivity of the film is, as
long as the dependence of the brightness of the negative depends in
a continuous manner on the brightness of the object being pho-
tographed, there will be a shade of grey that produces exactly the
same shade of grey on the negative when photographed. This is the
essence of Wheeler and Feynman's idea. Let us first be a bit more
precise and then a bit more general.
For simplicity let us suppose that the film is always a uniform
shade of grey (i.e. at any time the shade of grey does not vary by
location on the film). The possible shades of grey of the film can
then be represented by the (real) numbers from 0, representing pure
black, to 1, representing pure white.
Let us now distinguish various stages in the chronological order of
the life of the film (see Figure 0)

Figure 0

In stage S^ the film is young; it has just been placed in the camera
and is ready to be exposed. It is then exposed to the object that
comes out of the time machine. (That object in fact is a later stage
of the film itself). By the time we come to stage S2 of the life of the
171
Frank Arntzenius and Tim Maudlin

film, it has been developed and is about to enter the time machine.
Stage S 3 occurs just after it exits the time machine and just before
it is photographed. Stage S 4 occurs after it has been photographed
and before it starts fading away.
Let us assume that the film starts out, in stage Sj in some uni-
form shade of grey, and that the only significant change in the shade
of grey of the film occurs between stages Sj and S 2 During that
period it acquires a shade of grey that depends on the shade of grey
of the object that was photographed. I.e. the shade of grey that the
film acquires at stage S2 depends on the shade of grey it has at stage
S3. The influence of the shade of grey of the film at stage S3, on the
shade of grey of the film at stage S 2 , can be represented as a map-
ping, or function, from the real numbers between 0 and 1 (inclu-
sive), to the real numbers between 0 and 1 (inclusive). Let us sup-
pose that the process of photography is such that if one imagines
varying the shade of grey of an object in a smooth, continuous man-
ner then the shade of grey of the developed picture of that object
will also vary in a smooth, continuous manner. This implies that the
function in question will be a continuous function. Now any con-
tinuous function from the real numbers between 0 and 1 (inclusive)
to the real numbers between 0 and 1 (inclusive) must map at least
one number to itself. One can quickly convince oneself of this by
graphing such functions. For one will quickly see that any continu-
ous function f from [0,1] to [0,1] must intersect the line x=y some-
where, and thus there must be at least one point x such that f(x)=x.
Such points are called fixed points of the function. Now let us think
about what such a fixed point represents. It represents a shade of
grey such that, when photographed, it will produce a developed film
with exactly that same shade of grey. The existence of such a fixed
point implies a solution to the apparent paradox.
Let us now be more general and allow colour photography. One
can represent each possible colour of an object (of uniform colour)
by the proportions of blue, green and red that make up that colour.
(This is why television screens can produce all possible colours.)
Thus one can represent all possible colours of an object by three
points on three orthogonal lines x, y and z, that is to say, by a point
in a three-dimensional cube. This cube is also known as the
'Cartesian product' of the three line segments. Now, one can also
show that any continuous map from such a cube to itself must have
at least one fixed point. So colour photography can not be used to
create time travel paradoxes either!
Even more generally, consider some system P which, as in the
above example, has the following life. It starts in some state S 1; it
172
 Time Travel and Modern Physics

interacts with an object that comes out of a time machine (which

happens to be its older self), it travels back in time, it interacts with
some object (which happens to be its younger self), and finally it
grows old and dies. Let us assume that the set of possible states of
P can be represented by a Cartesian product of n closed intervals of
the reals, i.e. let us assume that the topology of the state-space of P
is isomorphic to a finite Cartesian product of closed intervals of the
reals. Let us further assume that the development of P in time, and
the dependence of that development on the state of objects that it
interacts with, is continuous. Then, by a well-known fixed point
theorem in topology (see e.g. Hocking and Young 1961, p. 273), no
matter what the nature of the interaction is, and no matter what the
initial state of the object is, there will be at least one state 5 3 of the
older system (as it emerges from the time travel machine) that will
influence the initial state S\ of the younger system (when it encoun-
ters the older system) so that, as the younger system becomes older,
it develops exactly into state £ 3 . Thus without imposing any con-
straints on the initial state 5j of the system P, we have shown that
there will always be perfectly ordinary, non-paradoxical, solutions,
in which everything that happens, happens according to the usual
laws of development. Of course, there is looped causation, hence
presumably also looped explanation, but what do you expect if there
is looped time?
Unfortunately, for the fan of time travel, a little reflection
suggests that there are systems for which the needed fixed point
theorem does not hold. Imagine, for instance, that we have a dial
that can only rotate in a plane. We are going to put the dial in the
time machine. Indeed we have decided that if we see the later stage
of the dial come out of the time machine set at angle x, then we will
set the dial to x+90, and throw it into the time machine. Now it
seems we have a paradox, since the mapping that consists of a rota-
tion of all points in a circular state-space by 90 degrees does not
have a fixed point. And why wouldn't some state-spaces have the
topology of a circle?
However, we have so far not used another continuity assumption
which is also a reasonable assumption. So far we have only made the
following demand: the state the dial is in at stage must S2 be a con-
tinuous function of the state of the dial at stage £3. But, the state of
the dial at stage 5 2 is arrived at by taking the state of the dial at stage
Sj and rotating it over some angle. It is not merely the case that the
effect of the interaction, namely the state of the dial at stage S2,
should be a continuous function of the cause, namely the state of
the dial at stage 5 3 . It is additionally the case that the path taken to
173
Frank Arntzenius and Tim Maudlin

get there, the way the dial is rotated between stages Sx and S2 must
be a continuous function of the state at stage <S3. And, rather sur-
prisingly, it turns out that this can not be done. Let us illustrate
what the problem is before going to a more general demonstration
that there must be a fixed point solution in the dial case.
Forget time travel for the moment. Suppose that you and I each
have a watch with a single dial neither of which is running. My
watch is set at 12. You are going to announce what your watch is set
at. My task is going to be to adjust my watch to yours no matter
what announcement you make. And my actions should have a con-
tinuous (single valued) dependence on the time that you announce.
Surprisingly, this is not possible! For instance, suppose that if you
announce '12', then I achieve that setting on my watch by doing
nothing. Now imagine slowly and continuously increasing the
announced times, starting at 12. By continuity, I must achieve each
of those settings by rotating my dial to the right. If at some point I
switch and achieve the announced goal by a rotation of my dial to
the left, I will have introduced a discontinuity in my actions, a dis-
continuity in the actions that I take as a function of the announced
angle. So I will be forced, by continuity, to achieve every announce-
ment by rotating the dial to the right. But, this rotation to the right
will have to be abruptly discontinued as the announcements grow
larger and I eventually approach 12 again, since I achieved 12 by not
rotating the dial at all. So, there will be a discontinuity at 12 at the
latest. In general, continuity of my actions as a function of
announced times can not be maintained throughout if I am to be
able to replicate all possible settings. Another way to see the prob-
lem is that one can similarly reason that, as one starts with 12, and
imagines continuously making the announced times earlier, one will
be forced, by continuity, to achieve the announced times by rotating
the dial to the left. But the conclusions drawn from the assumption
of continuous increases and the assumption of continuous decreas-
es are inconsistent. So we have an inconsistency following from the
assumption of continuity and the assumption that I always manage
to set my watch to your watch. So, a dial developing according to a
continuous dynamics from a given initial state, can not be set up so
as to react to a second dial, with which it interacts, in such a way that
it is guaranteed to always end up set at the same angle as the second
dial. Similarly, it can not be set up so that it is guaranteed to always
end up set at 90 degrees to the setting of the second dial. All of this
has nothing to do with time travel. However, the impossibility of
such set ups is what prevents us from enacting the rotation by 90
degrees that would create paradox in the time travel setting.
174
 Time Travel and Modern Physics

Let us now give the positive result that with such dials there will
always be fixed point solutions, as long as the dynamics is continu-
ous. Let us call the state of the dial before it interacts with its older
self the initial state of the dial. And let us call the state of the dial
after it emerges from the time machine the final state of the dial. We
can represent the possible initial and final states of the dial by the
angles x and y that the dial can point at initially and finally. The set
of possible initial plus final states thus forms a torus. (See figure 1.)

Angle X

Figure 1
Suppose that the dial starts at angle /. The initial angle I that the
dial is at before it encounters its older self, and the set of all possi-
ble final angles that the dial can have when it emerges from the time
machine is represented by the circle / on the torus (see figure 1).
Given any possible angle of the emerging dial the dial initially at
angle / will develop to some other angle. One can picture this devel-
opment by rotating each point on / in the horizontal direction by the
relevant amount. Since the rotation has to depend continuously on
the angle of the emerging dial, ring / during this development will
deform into some loop L on the torus. Loop L thus represents the
angle x that the dial is at when it is thrown into the time machine,
given that it started at angle / and then encountered a dial (its older
self) which was at angle y when it emerged from the time machine.
We therefore have consistency if x=y for some x and y on loop L.
Now, let loop C be the loop which consists of all the points on the
torus for which x-y. Ring / intersects C at point <i,i>. Obviously
any continuous deformation of / must still intersect C somewhere.
So L must intersect C somewhere, say at <jj>. But that means that
no matter how the development of the dial starting at / depends on
the angle of the emerging dial, there will be some angle for the
emerging dial such that the dial will develop exactly into that angle

175
Frank Arntzenius and Tim Maudlin

(by the time it enters the time machine) under the influence of that
emerging dial. This is so no matter what angle one starts with, and
no matter how the development depends on the angle of the emerg-
ing dial. Thus even for a circular state-space there are no constraints
needed other than continuity.
Unfortunately there are state-spaces that escape even this argu-
ment. Consider for instance a pointer that can be set to all values
between 0 and 1, where 0 and 1 are not possible values. I.e. suppose
that we have a state-space that is isomorphic to an open set of real
numbers. Now suppose that we have a machine that sets the point-
er to half the value that the pointer is set at when it emerges from
the time machine.

Suppose the pointer starts at value I. As before we can represent

the combination of this initial position and all possible final posi-
tions by the line /. Under the influence of pointer coming out of the
time machine the pointer value will develop to a value that equals
half the value of the final value that it encountered. We can repre-
sent this development as the continuous deformation of line / into
line L. This development is fully continuous. A point <x,y> on line
/ represents the initial position x of the (young) pointer and the
position y of the older pointer as it emerges from the time machine.
176
 Time Travel and Modern Physics

A point <x, y> on line L represents the value x that the pointer
should develop into given that it encountered the older pointer
emerging from the time machine set at position y. Since the point-
er is designed to develop to half the value of the pointer that it
encounters, the line L corresponds to x-\/2y. We have consistency
if there is some point such that it develops into that point, i.e. if
there is some point <x,y> on line L such that x=y. However, there
is no such point: lines L and C do not intersect. Thus there is no
consistent solution, despite the fact that the dynamics is fully con-
tinuous.
Of course if 0 were a possible value L and C would intersect at
<0,0>. This is surprising and strange: adding one point to the set of
possible values of a quantity here makes the difference between
paradox and peace. One might be tempted to just add the extra
point to the state-space in order to avoid problems. After all, one
might say, surely no measurements could ever tell us whether the set
of possible values includes that exact point or not. Unfortunately
there can be good theoretical reasons for supposing that some quan-
tity has a state-space that is open: the set of all possible speeds of
massive objects in special relativity surely is an open set, since it
includes all speeds up to, but not including, the speed of light.
Quantities that have possible values that are not bounded also lead
to counter examples to the presented fixed point argument. And it
is not obvious to us why one should exclude such possibilities. So
the argument that no constraints are needed is not fully general.
An interesting question of course is: exactly for which state-
spaces must there be such fixed points. We do not know the general
answer.

The General Possibility of Time Travel in General Relativity

Time travel has recently been discussed quite extensively in the

context of general relativity. Time travel can occur in general rela-
tivistic models in which one has closed time-like curves (CTC's).
Travelling along such a curve, one would never exceed the speed of
light, and yet after a certain amount of (proper) time return to a
point in space-time that one previously visited. Or, by staying close
to such a CTC, one could come arbitrarily close to a point in space-
time that one previously visited. General relativity, in a rather
straightforward sense, allows time travel: there appear to be many
space-times (compatible with the fundamental equations of General
Relativity) in which there are CTC's. Space-time, for instance,
177
Frank Arntzenius and Tim Maudlin

could have a Minkowski metric everywhere, and yet have CTC's

everywhere by having the temporal dimension (topologically) rolled
up as a circle. Or, one can have wormhole connections between dif-
ferent parts of space-time which allow one to enter 'mouth A' of
such a wormhole connection, travel through the wormhole, exit the
wormhole at 'mouth B' and re-enter 'mouth A' again. Or, one can
have space-times which topologically are R4, and yet have CTC's
due to the 'tilting' of light cones (Godel space-times, Taub-NUT
space- times, etc.)
General relativity thus appears to provide ample opportunity for
time travel. Note that just because there are CTC's in a space-time,
this does not mean that one can get from any point in the space-time
to any other point by following some future directed timelike curve.
In many space-times in which there are CTC's such CTC's do not
occur all over space-time. Some parts of space-time can have CTC's
while other parts do not. Let us call the part of a space-time that has
CTC's the 'time travel region' of that space-time, while calling the
rest of that space-time the 'normal region'. More precisely, the
'time travel region' consists of all the space-time points p such that
there exists a (non-zero length) timelike curve that starts at p and
returns to p. Now let us start examining space-times with CTC's a
bit more closely for potential problems.

Two Toy Models

In order to get a feeling for the sorts of implications that closed

timelike curves can have, it may be useful to consider two simple
models. In space-times with closed timelike curves the traditional
initial value problem cannot be framed in the usual way. For it pre-
supposes the existence of Cauchy surfaces, and if there are CTCs
then no Cauchy surface exists. (A Cauchy surface is a spacelike sur-
face such that every inextendible timelike curve crosses it exactly
once. One normally specifies initial conditions by giving the condi-
tions on such a surface.) Nonetheless, if the topological complexi-
ties of the manifold are appropriately localized, we can come quite
close. Let us call an edgeless spacelike surface S a quasi-Cauchy sur-
face if it divides the rest of the manifold into two parts such that a)
every point in the manifold can be connected by a timelike curve to
S, and b) any timelike curve which connects a point in one region to
a point in the other region intersects S exactly once. It is obvious
that a quasi-Cauchy surface must entirely inhabit the normal region
of the space-time; if any point p of S is in the time travel region,
178
 Time Travel and Modern Physics

then any timelike curve which intersects p can be extended to a

timelike curve which intersects S near p again. In extreme cases of
time travel, a model may have no normal region at all (e.g.
Minkowski space-time rolled up like a cylinder in a time-like direc-
tion), in which case our usual notions of temporal precedence will
not apply. But temporal anomalies like wormholes (and time
machines) can be sufficiently localized to permit the existence of
quasi-Cauchy surfaces.
Given a timelike orientation, a quasi-Cauchy surface unproblem-
atically divides the manifold into its past (i.e. all points that can be
reached by past-directed timelike curves from S) and its future
(ditto mutatis mutandis). If the whole past of S is in the normal
region of the manifold, then S is a partial Cauchy surface: every
inextendible timelike curve which exists to the past of S intersects
S exactly once, but (if there is time travel in the future) not every
inextendible timelike curve which exists to the future of S intersects
S. Now we can ask a particularly clear question: consider a manifold
which contains a time travel region, but also has a partial Cauchy
surface S, such that all of the temporal funny business is to the
future of S. If all you could see were S and its past, you would not
know that the space-time had any time travel at all. The question is:
are there any constraints on the sort of data which can be put on S
and continued to a global solution of the dynamics which are dif-
ferent from the constraints (if any) on the data which can be put on
a Cauchy surface in a simply connected manifold and continued to
a global solution?
It is not at all surprising that there might be constraints on the
data which can be put on a locally space-like surface which passes
through the time travel region: after all, we never think we can
freely specify what happens on a space-like surface and on another
such surface to its future, but in this case the surface at issue lies to
its own future. But if there were particular constraints for data on a
partial Cauchy surface then we would apparently need to have to
rule out some sorts of otherwise acceptable states on S if there is to
be time travel to the future of S. We then might be able to establish
that there will be no time travel in the future by simple inspection
of the present state of the universe. As we will see, there is reason
to suspect that such constraints on the partial Cauchy surface are
non-generic. But we are getting ahead of ourselves: first let's con-
sider the effect of time travel on a very simple dynamics.
The simplest possible example is the Newtonian theory of
perfectly elastic collisions among equally massive particles in one
spatial dimension. The space-time is two-dimensional, so we can
179
Frank Arntzenius and Tim Maudlin

represent it initially as the Euclidean plane, and the dynamics is

completely specified by two conditions. When particles are travel-
ling freely, their world lines are straight lines in the space-time, and
when two particles collide, they exchange momenta, so the collision
looks like an 'X' in space-time, with each particle changing its
momentum at the impact.1 The dynamics is purely local, in that one
can check that a set of world-lines constitutes a model of the
dynamics by checking that the dynamics is obeyed in every arbi-
trarily small region. It is also trivial to generate solutions from arbi-
trary initial data if there are no CTCs: given the initial positions and
momenta of a set of particles, one simply draws a straight line from
each particle in the appropriate direction and continues it indefi-
nitely. Once all the lines are drawn, the worldline of each particle
can be traced from collision to collision. The boundary value prob-
lem for this dynamics is obviously well-posed: any set of data at an
instant yields a unique global solution, constructed by the method
sketched above.
What happens if we change the topology of the space-time by
hand to produce CTCs? The simplest way to do this is depicted in
figure 3: we cut and paste the space-time so it is no longer simply
connected by identifying the line L- with the line L+. Particles
'going in' to L+ from below 'emerge' from L-, and particles 'going
in' to L— from below 'emerge' from L+.

Figure 3. Inserting CTCs by Cut and Paste

How is the boundary-value problem changed by this alteration in

the space-time? Before the cut and paste, we can put arbitrary data
1
Multiple collisions are handled in the obvious way by continuity con-
siderations: just continue straight lines through the collision point and
identify which particle is which by their ordering in space.
180
 Time Travel and Modern Physics

on the simultaneity slice 5 and continue it to a unique solution.

After the change in topology, S is no longer a Cauchy surface, since
a CTC will never intersect it, but it is a partial Cauchy surface. So
we can ask two questions. First, can arbitrary data on 5 always be
continued to a global solution? Second, is that solution unique? If
the answer to the first question is no, then we have a backward-tem-
poral constraint: the existence of the region with CTCs places con-
straints on what can happen on S even though that region lies com-
pletely to the future of S. If the answer to the second question is no,
then we have an odd sort of indeterminism: the complete physical
state on 5 does not determine the physical state in the future, even
though the local dynamics is perfectly deterministic and even
though there is no other past edge to the space-time region in S's
future (i.e. there is nowhere else for boundary values to come from
which could influence the state of the region).
In this case the answer to the first question is yes and to the sec-
ond is no: there are no constraints on the data which can be put on
S, but those data are always consistent with an infinitude of differ-
ent global solutions. The easy way to see that there always is a solu-
tion is to construct the minimal solution in the following way. Start
drawing straight lines from S as required by the initial data. If a line
hits L- from the bottom, just continue it coming out of the top of
L+ in the appropriate place, and if a line hits L+ from the bottom,
continue it emerging from L- at the appropriate place. Figure 4 rep-
resents the minimal solution for a single particle which enters the
time-travel region from the left:

Figure 4. The Minimal Solution

The particle 'travels back in time' three times. It is obvious that

this minimal solution is a global solution, since the particle always
travels inertially.
181
Frank Arntzenius and Tim Maudlin

But the same initial state on 5 is also consistent with other glob-
al solutions. The new requirement imposed by the topology is just
that the data going into L+ from the bottom match the data coming
out of L- from the top, and the data going into L- from the bottom
match the data coming out of L+ from the top. So we can add any
number of vertical lines connecting L- and L+ to a solution and still
have a solution. For example, adding a few such lines to the mini-
mal solution yields:

L-

Figure 5. A Non-Minimal Solution

The particle now collides with itself twice: first before it reaches
L+ for the first time, and again shortly before it exits the CTC
region. From the particle's point of view, it is travelling to the right
at a constant speed until it hits an older version of itself and comes
to rest. It remains at rest until it is hit from the right by a younger
version of itself, and then continues moving off, and the same
process repeats later. It is clear that this is a global model of the
dynamics, and that any number of distinct models could be gener-
ating by varying the number and placement of vertical lines.
Knowing the data on S, then, gives us only incomplete informa-
tion about how things will go for the particle. We know that the par-
ticle will enter the CTC region, and will reach L+, we know that it
will be the only particle in the universe, we know exactly where and
with what speed it will exit the CTC region. But we cannot deter-
mine how many collisions the particle will undergo (if any), nor
how long (in proper time) it will stay in the CTC region. If the par-
ticle were a clock, we could not predict what time it would indicate
when exiting the region. Furthermore, the dynamics gives us no

182
 Time Travel and Modern Physics

handle on what to think of the various possibilities: there are no

probabilities assigned to the various distinct possible outcomes.
Changing the topology has changed the mathematics of the situ-
ation in two ways, which tend to pull in opposite directions. On the
one hand, S is no longer a Cauchy surface, so it is perhaps not sur-
prising that data on S do not suffice to fix a unique global solution.
But on the other hand, there is an added constraint: data 'coming
out' of L- must exactly match data 'going in' to L+, even though
what comes out of L- helps to determine what goes into L+. This
added consistency constraint tends to cut down on solutions,
although in this case the additional constraint is more than out-
weighed by the freedom to consider various sorts of data on L+/L-.
The fact that the extra freedom outweighs the extra constraint
also points up one unexpected way that the supposed paradoxes of
time travel may be overcome. Let's try to set up a paradoxical
situation using the little closed time loop above. If we send a single
particle into the loop from the left and do nothing else, we know
exactly where it will exit the right side of the time travel region.
Now suppose we station someone at the other side of the region
with the following charge: if the particle should come out on the
right side, the person is to do something to prevent the particle from
going in on the left in the first place. In fact, this is quite easy to do:
if we send a particle in from the right, it seems that it can exit on
the left and deflect the incoming left-hand particle.
Carrying on our reflection in this way, we further realize that if
the particle comes out on the right, we might as well send it back in
order to deflect itself from entering in the first place. So all we really
need to do is the following: set up a perfectly reflecting particle
mirror on the right-hand side of the time travel region, and launch
the particle from the left so that—if nothing interferes with it—it will
just barely hit L+. Our paradox is now apparently complete. If, on
the one hand, nothing interferes with the particle it will enter the
time-travel region on the left, exit on the right, be reflected from the
mirror, re-enter from the right, and come out on the left to prevent
itself from ever entering. So if it enters, it gets deflected and never
enters. On the other hand, if it never enters then nothing goes in on
the left, so nothing comes out on the right, so nothing is reflected
back, and there is nothing to deflect it from entering. So if it doesn't
enter, then there is nothing to deflect it and it enters. If it enters,
then it is deflected and doesn't enter; if it doesn't enter then there is
nothing to deflect it and it enters: paradox complete.
But at least one solution to the supposed paradox is easy to con-
struct: just follow the recipe for constructing the minimal solution,
183
Frank Arntzenius and Tim Maudlin

continuing the initial trajectory of the particle (reflecting it the

mirror in the obvious way) and then read off the number and tra-
jectories of the particles from the resulting diagram. We get the
result of figure 6:

L-

S mrror

Figure 6. Resolving the 'Paradox'

As we can see, the particle approaching from the left never reaches
L+: it is deflected first by a particle which emerges from L-. But it
is not deflected by itself, as the paradox suggests, it is deflected by
another particle. Indeed, there are now four particles in the diagram:
the original particle and three particles which are confined to closed
time-like curves. It is not the leftmost particle which is reflected by
the mirror, nor even the particle which deflects the leftmost
particle; it is another particle altogether.
The paradox gets it traction from an incorrect presupposition: if
there is only one particle in the world at S then there is only one
particle which could participate in an interaction in the time travel
region: the single particle would have to interact with its earlier (or
later) self. But there is no telling what might come out of L-: the
only requirement is that whatever comes out must match what goes
in at L+. So if you go to the trouble of constructing a working time
machine, you should be prepared for a different kind of disappoint-
ment when you attempt to go back and kill yourself: you may be
prevented from entering the machine in the first place by some
completely unpredictable entity which emerges from it. And once
again a peculiar sort of indeterminism appears: if there are many
self-consistent things which could prevent you from entering, there
is no telling which is even likely to materialize.
184
 Time Travel and Modern Physics

So when the freedom to put data on L- outweighs the constraint

that the same data go into L+, instead of paradox we get an embar-
rassment of riches: many solution consistent with the data on S. To
see a case where the constraint 'outweighs' the freedom, we need to
construct a very particular, and frankly artificial, dynamics and
topology. Consider the space of all linear dynamics for a scalar field
on a lattice. We will depict the space-time lattice as a directed graph.
There is to be a scalar field defined at every node of the graph,
whose value at a given node depends linearly on the values of the
field at nodes which have arrows which lead to it. Each edge of the
graph can be assigned a weighting factor which determines how
much the field at the input node contributes to the field at the out-
put node. If we name the nodes by the letters a, b, c, etc., and the
edges by their endpoints in the obvious way, then we can label the
weighting factors by the edges they are associated with in an
equally obvious way.
Suppose that the graph of the space-time lattice is acyclic, as in
figure 7.

/VV
Figure 7. An Acyclic Lattice

It is easy to regard a set of nodes as the analog of a Cauchy

surface, e.g. the set {a, b, c}, and it is obvious if arbitrary data are
put on those nodes the data will generate a unique solution in the
185
Frank Arntzenius and Tim Maudlin

future.2 If the value of the field at node a is 3 and at node b is 7, then

its value at node d will be 3Wad and its value at node e will be 3Wae
+ 7W6e. By varying the weighting factors we can adjust the dynam-
ics, but in an acyclic graph the future evolution of the field will
always be unique.
Let us now again artificially alter the topology of the lattice to
admit CTCs. One of the simplest such graphs is depicted in figure
8: there are now paths which lead from z back to itself.

Figure 8. Time Travel on a Lattice

Can we now put arbitrary data on v and w, and continue that data
to a global solution? Will the solution be unique?
In the generic case, there will be a solution and the solution will
be unique. The equations for the value of the field at x, y, and z are:
x = vWvx
y - wWvy y
z = xWxz + yWyz.
Solving these equations for z yields
* = (vWra + zWzx)Wxz
or
z = (vWMW
which gives a unique value for z in the generic case. But looking at
the space of all possible dynamics for this lattice (i.e. the space of all
possible weighting factors), we find a singularity in the case where
1 - yVzxWxz - W2vWj,2 = 0. If we choose weighting factors in just
2
The dynamics here is radically non-time-reversible. Indeed, the dynam-
ics is deterministic in the future direction but not in the past direction.
186
 Time Travel and Modern Physics

this way, then arbitrary data at v and w cannot be continued to a

global solution. Indeed, if the scalar field is everywhere non-nega-
tive, then this particular choice of dynamics puts ironclad con-
straints on the value of the field at v and w: the field there must be
zero (assuming Wv;c and WWJ, to be non-zero), and similarly all
nodes in their past must have field value zero. If the field can take
negative values, then the values at v and w must be so chosen that
vWvxWxz = -wWwyWyz. In either case, the field values at v and w
are severely constrained by the existence of the CTC region even
though these nodes lie completely to the past of that region. It is
this sort of constraint which we find to be unlike anything which
appears in standard physics.
Our toy models suggest three things. The first is that it may be
impossible to prove in complete generality that arbitrary data on a
partial Cauchy surface can always be continued to a global solution:
our artificial case provides an example where it cannot. The second
is that such odd constraints are not likely to be generic: we had to
delicately fine-tune the dynamics to get a problem. The third is that
the opposite problem, namely data on a partial Cauchy surface
being consistent with many different global solutions, is likely to be
generic: we did not have to do any fine-tuning to get this result. And
this leads to a peculiar sort of indeterminism: the entire state on 5
does not determine what will happen in the future even though the
local dynamics is deterministic and there are no other 'edges' to
space-time from which data could influence the result. What hap-
pens in the time travel region is constrained but not determined by
what happens on S, and the dynamics does not even supply any
probabilities for the various possibilities. The example of the photo-
graphic negative discussed in section 3, then, seems likely to be
unusual, for in that case there is a unique fixed point for the dynam-
ics, and the set-up plus the dynamical laws determine the outcome.
In the generic case one would rather expect multiple fixed points,
with no room for anything to influence, even probabilistically, which
would be realized.
It is ironic that time travel should lead generically not to contra-
dictions or to constraints (in the normal region) but to underdeter-
mination of what happens in the time travel region by what happens
everywhere else (an underdetermination tied neither to a proba-
bilistic dynamics or to a free edge to space-time). The traditional
objection to time travel is that it leads to contradictions: there is no
consistent way to complete an arbitrarily constructed story about
how the time traveler intends to act. Instead, though, it appears that
the problem is underdetermination: the story can be consistently
187
Frank Arntzenius and Tim Maudlin

completed in many different ways. Let us now discuss some results

regarding some slightly more realistic models that have been dis-
cussed in the physics literature.

Slightly More Realistic Models of Time Travel

Echeverria, Klinkhammer and Thorne (1991) considered the case of

3-dimensional single hard spherical ball that can go through a
single time travel wormhole so as to collide with its younger self.
The threat of paradox in this case arises in the following form.
There are initial trajectories for the ball such that if such a trajectory

l Figure 9
is continued assuming that the ball does not undergo a collision
prior to entering mouth 1 of the wormhole, it will exit mouth 2 so
as to collide with its earlier self prior to its entry into mouth 1 in
such a way as to prevent its earlier self from entering mouth 1. Thus
it seems that the ball will enter mouth 1 if and only if it does not
enter mouth 1. Of course, the Wheeler-Feynman strategy is to look
for a 'glancing blow' solution: a collision which will produce exactly
the (small) deviation in trajectory of the earlier ball that produces
exactly that collision. Are there always such solutions?3
3
One might hope that fixed point theorems can be used to prove the
existence of solutions in this type of cases too. Consider, for instance, a
fixed initial state of motion I of the ball. Then consider all the possible
velocities and locations and times at which such a ball could enter mouth
1 of the wormhole. Each such triple will determine the trajectory of that
ball out of mouth 2. One can then look at the continuation of the
trajectory from state I and that from state s, and see whether these trajec-
tories collide. Then one can see for each possible triple whether the ball
that starts in state I will be collided into mouth 1, and if it is, with which
speed at what location and at which time this will occur. Thus given state
I, each triple maps onto another triple <v',x',t'>. One might then suggest
188
 Time Travel and Modern Physics

Echeverria, Klinkhammer & Thorne found a large class of initial

trajectories that have consistent 'glancing blow' continuations, and
found none that do not (but their search was not completely general).
They did not produce a rigorous proof that every initial trajectory
has a consistent continuation, but suggested that it is very plausible
that every initial trajectory has a consistent continuation.
In fact, as one might expect from our discussion in the previous
section, they found the opposite problem from that of inconsistency:
they found underdetermination. For a large class of initial trajecto-
ries there are multiple different consistent 'glancing blow' continu-
ations of that trajectory (many of which involve multiple wormhole
traversals). For example, if one initially has a ball that is travelling
on a trajectory aimed straight between the two mouths, then one
obvious solution is that the ball passes between the two mouths and
never time travels. But another solution is that the younger ball gets
knocked into mouth 1 exactly so as to come out of mouth 2 and pro-
duce that collision. Echeverria et. al. do not note the possibility
(which we pointed out in the previous section) of the existence of
additional balls in the time travel region. We conjecture (but have
no proof) that for every initial trajectory of A there are some, and
generically many, multiple ball continuations.
Friedman et. al. 1990 examined the case of source free non-self-
interacting scalar fields travelling through such a time travel worm-
hole and found that no constraints on initial conditions are imposed
by the existence of these time travel wormholes. In general there
appear to be no known counter examples to the claim that in 'some-
what realistic' time-travel space-times with a partial Cauchy surface
there are no constraints imposed on the state on such a partial Cauchy
surface by the existence of CTC's. (See e.g. Friedman and Morris
1991, Thorne 1994 Earman 1995, Earman and Smeenk 1999.)
How about the issue of constraints in the time travel region?
Prima facie, constraints in such a region would not appear to be sur-

appealing to a fixed point theorem to argue that there must be a solution

for each initial state I. However, in the first place the set of possible speeds
and times are open sets. And in the second place there can be multiple
wormhole traversals. Thus the relevant total state-space of wormhole
mouth crossings consists of discretely many completely disconnected
state-spaces (with increasing numbers of dimensions). So standard fixed
point theorems do not apply directly. It should be noted that the results
that have been achieved regarding this case do make use of fixed points
theorems quite extensively. But their application is limited to certain sub-
problems, and do not yield a fully general proof of the lack of constraints
for arbitrary I.
189
Frank Arntzenius and Tim Maudlin

prising. But one might still expect that there should be no con-
straints on states on a spacelike surface, provided one keeps the sur-
face 'small enough'. In the physics literature the following question
has been asked: for any point p in T, and any space-like surface S
that includes p is there a neighbourhood E of p in S such that any
solution on E can be extended to a solution on the whole space-
time? With respect to this question, there are some simple models
in which one has this kind of extendibility of local solutions to glob-
al ones, and some simple models in which one does not have such
extendibility, with no clear general pattern. (See e.g. Yurtsever
1990, Friedman et. al. 1990, Novikov 1992, Earman 1995, Earman
and Smeenk 1999). What are we to think of all of this?

Even If There are Constraints, So What?

Since it is not obvious that one can rid oneself of all constraints in
realistic models, let us examine the argument that time travel is
implausible, and we should think it unlikely to exist in our world, in
so far as it implies such constraints. The argument goes something
like the following. In order to satisfy such constraints one needs
some pre-established divine harmony between the global (time trav-
el) structure of space-time and the distribution of particles and
fields on space-like surfaces in it. But it is not plausible that the
actual world, or any world even remotely like ours, is constructed
with divine harmony as part of the plan. In fact, one might argue,
we have empirical evidence that conditions in any spatial region can
vary quite arbitrarily. So we have evidence that such constraints,
whatever they are, do not in fact exist in our world. So we have evi-
dence that there are no closed time-like lines in our world or one
remotely like it. We will now examine this argument in more detail
by presenting four possible responses, with counterresponses, to
this argument.

Response 1. There is nothing implausible or new about such con-

straints. For instance, if the universe is spatially closed, there has to
be enough matter to produce the needed curvature, and this puts
constraints on the matter distribution on a space-like hypersurface.
Thus global space-time structure can quite unproblematically con-
strain matter distributions on space-like hypersurfaces in it.
Moreover we have no realistic idea what these constraints look like,
so we hardly can be said to have evidence that they do not obtain.
Counterresponse 1. Of course there are constraining relations
190
 Time Travel and Modern Physics

between the global structure of space-time and the matter in it. The
Einstein equations relate curvature of the manifold to the matter
distribution in it. But what is so strange and implausible about the
constraints imposed by the existence of closed time-like curves is
that these constraints in essence have nothing to do with the
Einstein equations. When investigating such constraints one typi-
cally treats the particles and/or field in question as test particles
and/or fields in a given space-time, i.e. they are assumed not to
affect the metric of space-time in any way. In typical space-times
without closed time-like curves this means that one has, in essence,
complete freedom of matter distribution on a space-like hypersur-
face. (See response 2 for some more discussion of this issue). The
constraints imposed by the possibility of time travel have a quite
different origin and are implausible. In the ordinary case there is a
causal interaction between matter and space-time that results in
relations between global structure of space-time and the matter dis-
tribution in it. In the time travel case there is no such causal story
to be told: there simply has to be some pre-established harmony
between the global space-time structure and the matter distribution
on some space-like surfaces. This is implausible.

Response 2. Constraints upon matter distributions are nothing new:

for instance 2?=div(p) constrains field/particle distributions on
space-like hypersurfaces. This is not implausible divine harmony.
Such constraints can hold as a matter of physical law. Moreover, if
we had inferred from the apparent free variation of conditions on
spatial regions that there could be no such constraints we would
have mistakenly inferred that i?=div(p) could not be a law of nature.

Counterresponse 2. The constraints imposed by the existence of

closed time-like lines are of quite a different character from the con-
straint imposed by £'=div(p). The constraints imposed by i?=div(p)
on the state on a space-like hypersurface are i) local constraints (i.e.
to check whether the constraint holds in a region you just need to
see whether it holds at each point in the region), ii) quite indepen-
dent of the global space-time structure, iii) quite independent of
how the space-like surface in question is embedded in a given space-
time, and iv) very simply and generally stateable. On the other
hand, the consistency constraints imposed by the existence of
closed time-like curves i) are not local, ii) are dependent on the
global structure of space-time, iii) depend on the location of the
space-like surface in question in a given space-time, and iv) appear
not to be simply stateable other than as the demand that the state on
191
Frank Arntzenius and Tim Maudlin

that space-like surface embedded in such and such a way in a given

space-time, do not lead to inconsistency. On some views of laws
(e.g. David Lewis' view) this plausibly implies that such constraints,
even if they hold, could not possibly be laws. But even if one does
not accept such a view of laws, one could claim that the bizarre fea-
tures of such constraints imply that it is implausible that such con-
straints hold in our world or in any world remotely like ours.

Response 3. It would be strange if there are constraints in the non-

time travel region. It is not strange if there are constraints in the
time travel region. They should be explained in terms of the
strange, self-interactive, character of time travel regions. In this
region there are time-like trajectories from points to themselves.
Thus the state at such a point, in such a region, will, in a sense,
interact with itself. It is a well-known fact that systems that interact
with themselves will develop into an equilibrium state, if there is
such an equilibrium state, or else will develop towards some singu-
larity. Normally, of course, self-interaction isn't true instantaneous
self-interaction, but consists of a feed-back mechanism that takes
time. But in time travel regions something like true instantaneous
self-interaction occurs. This explains why constraints on states
occur in such time travel regions: the states 'ab initio' have to be
'equilibrium states'. Indeed in a way this also provides some picture
of why indeterminism occurs in time travel regions: at the onset of
self-interaction states can fork into different equi-possible equilib-
rium states.

Counterresponse 3. This is explanation by woolly analogy. It all

goes to show that time travel leads to such bizarre consequences that
it is unlikely that it occurs in a world remotely like ours.

Response 4. All of the previous discussion completely misses the

point. So far we have been taking the space-time structure as given,
and asked the question whether a given time travel space-time
structure imposes constraints on states on (parts of) space-like sur-
faces. However, space-time and matter interact. Suppose that one is
in a space-time with closed time-like lines, such that certain coun-
terfactual distributions of matter on some neighborhood of a point
p are ruled out if one holds that space-time structure fixed. One
might then ask 'Why does the actual state near p in fact satisfy these
constraints? By what divine luck or plan is this local state compati-
ble with the global space-time structure? What if conditions near p
had been slightly different?'. And one might take it that the lack of
192
 Time Travel and Modern Physics

normal answers to these questions indicates that it is very

implausible that our world, or any remotely like it, is such a time
travel universe. However the proper response to these question is
the following. There are no constraints in any significant sense. If
they hold they hold as a matter of accidental fact, not of law. There
is no more explanation of them possible than there is of any con-
tingent fact. Had conditions in a neighborhood of p been otherwise,
the global structure of space-time would have been different. So
what? The only question relevant to the issue of constraints is
whether an arbitrary state on an arbitrary spatial surface 5 can
always be embedded into a space-time such that that state on 5
consistently extends to a solution on the entire space-time.
But we know the answer to that question. A well-known theorem
in general relativity says the following: any initial data set on a three
dimensional manifold 5 with positive definite metric has a unique
embedding into a maximal space-time in which 5 is a Cauchy sur-
face (see e.g. Geroch and Horowitz 1979, p 284 for more detail), i.e.
there is a unique largest space-time which has S as a Cauchy surface
and contains a consistent evolution of the initial value data on 5.
Now since S is a Cauchy surface this space-time does not have
closed time like curves. But it may have extensions (in which S is
not a Cauchy surface) which include closed timelike curves, indeed
it may be that any maximal extension of it would include closed
timelike curves. (This appears to be the case for extensions of states
on certain surfaces of Taub-NUT space-times. See Earman and
Smeenk 1999). But these extensions, of course, will be consistent.
So properly speaking, there are no constraints on states on space-
like surfaces. Nonetheless the space-time in which these are embed-
ded may or may not include closed time-like curves.

Counterresponse 4. This, in essence, is the stonewalling answer

which we indicated at the beginning of section 2. However, whether
or not you call the constraints imposed by a given space-time on dis-
tributions of matter on certain space-like surfaces 'genuine con-
straints', whether or not they can be considered lawlike, and
whether or not they need to be explained, the existence of such con-
straints can still be used to argue that time travel worlds are so
bizarre that it is implausible that our world or any world remotely
like ours is a time travel world.
Suppose that one is in a time travel world. Suppose that given the
global space-time structure of this world, there are constraints
imposed upon, say, the state of motion of a ball on some space-like
surface when it is treated as a test particle, i.e. when it is assumed
193
Frank Arntzenius and Tim Maudlin

that the ball does not affect the metric properties of the space-time
it is in. (There is lots of other matter that, via the Einstein equation,
corresponds exactly to the curvature that there is everywhere in this
time travel worlds.) Now a real ball of course does have some effect
on the metric of the space-time it is in. But let us consider a ball that
is so small that its effect on the metric is negligible. Presumably it
will still be the case that certain states of this ball on that space-like
surface are not compatible with the global time travel structure of
this universe.
This means that the actual distribution of matter on such a space-
like surface can be extended into a space-time with closed time-like
lines, but that certain counterfactual distributions of matter on this
space-like surface can not be extended into the same space-time.
But note that the changes made in the matter distribution (when going
from the actual to the counterfactual distribution) do not in any non-
negligible way affect the metric properties of the space-time. Thus the
reason why the global time travel properties of the counterfactual
space-time have to be significantly different from the actual space-
time is not that there are problems with metric singularities or alter-
ations in the metric that force significant global changes when we go
to the counterfactual matter distribution. The reason that the coun-
terfactual space-time has to be different is that in the counterfactu-
al world the ball's initial state of motion starting on the space-like
surface, could not 'meet up' in a consistent way with its earlier self
(could not be consistently extended) if we were to let the global
structure of the counterfactual space-time be the same as that of the
actual space-time. Now, it is not bizarre or implausible that there is
a counterfactual dependence of manifold structure, even of its
topology, on matter distributions on spacelike surfaces. For
instance, certain matter distributions may lead to singularities, oth-
ers may not. We may indeed in some sense have causal power over
the topology of the space-time we live in. But this power normally
comes via the Einstein equations. But it is bizarre to think that there
could be a counterfactual dependence of global space-time struc-
ture on the arrangement of certain tiny bits of matter on some
space-like surface, where changes in that arrangement by assump-
tion do not affect the metric anywhere in space-time in any significant
way. It is implausible that we live in such a world, or that a world
even remotely like ours is like that.
Let us illustrate this argument in a different way by assuming
that wormhole time travel imposes constraints upon the states of
people prior to such time travel, where the people have so little
mass/energy that they have negligible effect, via the Einstein equa-
194
 Time Travel and Modern Physics

tion, on the local metric properties of space-time. Do you think it

more plausible that we live in a world where wormhole time travel
occurs but it only occurs when people's states are such that these
local states happen to combine with time travel in such a way that
nobody ever succeeds in killing their younger self, or do you think
it more plausible that we are not in a wormhole time travel world?4

Quantum Mechanics to the Rescue?

There has been a particularly clear treatment of time travel in the

context of quantum mechanics by David Deutsch (see Deutsch
1991, and Deutsch and Lockwood 1994) in which it is claimed that
quantum mechanical considerations show that time travel never
imposes any constraints on the pre-time travel state of systems. The
essence of this account is as follows.
A quantum system starts in state 51, interacts with its older self,
after the interaction is in state S2, time travels while developing into
state 5 3 , then interacts with its younger self, and ends in state 5 4
(see figure 10).

Figure 10

4
This argument, especially the second illustration of it, is similar to the
one in Horwich 1987, pp 124—8. However, we do not share Horwich's view
that it only tells against time travel of humans into their local past.
195
Frank Arntzenius and Tim Maudlin

Deutsch assumes that the set of possible states of this system are
the mixed states, i.e. are represented by the density matrices over
the Hilbert space of that system. Deutsch then shows that for any
initial state S^, any unitary interaction between the older and
younger self, and any unitary development during time travel, there
is a consistent solution, i.e. there is at least one pair of states £ 2 and
AS3 such that when S^ interacts with £3 it will change to state 5 2 and
S 2 will then develop into S 3 . The states S2, S 3 and £4 will typically
be not be pure states, i.e. will be non-trivial mixed states, even if Sj
is pure. In order to understand how this leads to interpretational
problems let us give an example. Consider a system that has a two
dimensional Hilbert space with as a basis the states |+> and |->.
Let us suppose that when state | +> of the young system encounters
state I +> of the older system, they interact and the young system
develops into state |—> and the old system remains in state | +>. In
obvious notation:
I +>j I +>3 develops into |-> 2 1 + > 4-
Similarly, suppose that:
I +>j I -> 3 develops into | +>21 +>4,
I ->j I +>3 develops into |-> 2 1->4 , and
|->j|-> 3 develops into |+> 2 |->4.
Let us furthermore assume that there is no development of the state
of the system during time travel, i.e. that | +> 2 develops into | +>3,
and that |-> 2 develops into |-> 3 .
Now, if the only possible states of the system were |+> and |->
(i.e. if there were no superpositions or mixtures of these states),
then there is a paradox. There is no initial state compatible with this
dynamics. For if | +>j interacts with | +>3 then it will develop into
|-> 2 , which, during time travel, will develop into |-> 3 , which
inconsistent with the assumed state | +>3. Similarly if | +>j inter-
acts with |-> 3 it will develop into |+> 2 , which will then develop
into I +>3 which is also inconsistent. Thus the system can not start
in state | +i>. Similar reasoning shows that one can not start in state
I ->j. So there is no initial state consistent with this dynamics.
But, says Deutsch, in quantum mechanics such a system can also
be in any mixture of the states | +> and |->. Suppose that the older
system, prior to the interaction, is in a state S 3 which is an equal
mixture of 50% | +> 3 and 50% |-> 3 . Then the younger system dur-
ing the interaction will develop into a mixture of 50% | +>2 and 50%
|-> 2 , which will then develop into a mixture of 50% | +>3 and 50%
|-> 3 , which is consistent! More generally Deutsch uses a fixed
196
 Time Travel and Modern Physics

point theorem to show that no matter what the unitary development

during interaction is, and no matter what the unitary development
during time travel is, for any state «Sj there is always a state £3
(which typically is not a pure state) which causes S^ to develop into
a state S2 which develops into that state S3. Thus quantum mechan-
ics comes to the rescue: it shows in all generality that no constraints
on initial states are needed!
One might wonder why Deutsch appeals to mixed states: will
superpositions of states |+> and | - > not suffice? Unfortunately
such an idea does not work. For instance, let us assume that state S3
is the superposition l/V2|+>3+l/v2|-> 3 . One might then wish to
claim that initial state | +>j when it encounters
1/V21+>3+l/v21->3, will develop into superposition
l/V2|+> 2 +l/V2|->2, and that this in turn will develop into
1/\2|+>3+1/"V2|—> 3 , as desired. However this is not correct. For
initial state | +>j when it encounters l/\21 +> 3 +l/v21-> 3 , will sub-
sequently develop into the entangled state
1 / V 2 | - > 2 | + > 4 + l / v 2 | + > 2 | - > 4 . In so far as one can speak of the
state of the young system after this interaction, it is in the mixture
of 50% |+> 2 and 50% |-> 2 , not in the superposition l/V2|+> 2 +
1/V2|—>2. So Deutsch does need his recourse to mixed states.
This clarification of why Deutsch needs his mixtures does how-
ever indicate a serious worry about the simplifications that are part
of Deutsch's account. After the interaction the old and young sys-
tem will (typically) be in an entangled state. Although for purposes
of a measurement on one of the two systems one can say that this
system is in a mixed state, one can not represent the full state of the
two systems by specifying the mixed state of each separate part, as
there are correlations between observables of the two systems that
are not represented by these two mixed states, but are represented
in the joint entangled state. But if there really is an entangled state
of the old and young systems directly after the interaction, how is
one to represent the subsequent development of this entangled
state? Will the state of the younger system remain entangled with
the state of the older system as the younger system time travels and
the older system moves on into the future? On what space-like sur-
faces are we to imagine this total entangled state to be? At this point
it becomes clear that there is no obvious and simple way to extend
elementary non-relativistic quantum mechanics to space-times with
closed time-like curves. There have been more sophisticated
approaches than Deutsch's to time travel, using technical machin-
ery from quantum field theory and differentiable manifolds (see e.g.
Friedman et al. 1991, Earman and Smeenk 1999, and references

197
Frank Arntzenius and Tim Maudlin

therein). But out of such approaches no results anywhere near as

clear and interesting as Deutsch's have been forthcoming.
How does Deutsch avoid these complications? Deutsch assumes
a mixed state 5 3 of the older system prior to the interaction with the
younger system. He lets it interact with an arbitrary pure state S]
younger system. After this interaction there is an entangled state S'
of the two systems. Deutsch computes the mixed state S2 of the
younger system which is implied by this entangled state S'. His
demand for consistency then is just that this mixed state S2 devel-
ops into the mixed state S 3 . Now it is not at all clear that this is a
legitimate way to simplify the problem of time travel in quantum
mechanics. But even if we grant him this simplification there is a
problem: how are we to understand these mixtures?
If we take an ignorance interpretation of mixtures we run into
trouble. For suppose that we assume that in each individual case
each older system is either in state | +>3 or in state |-> 3 prior to the
interaction. Then we regain our paradox. Deutsch instead recom-
mends the following, many worlds, picture of mixtures. Suppose we
start with state | +>j in all worlds. In some of the many worlds the
older system will be in the | +> 3 state, let us call them A -worlds, and
in some worlds, -B-worlds, it will be in the — | >3 state. Thus in A-
worlds after interaction we will have state |->2, and in .B-worlds we
will have state |+>2- During time travel the |-> 2 state will remain
the same, i.e turn into state —| >3, but the systems in question will
travel from A -worlds to B-worlds. Similarly the |+>2 states will
travel from the B-worlds to the A -worlds, thus preserving consis-
tency.
Now whatever one thinks of the merits of many worlds interpre-
tations, and of this understanding of it applied to mixtures, in the
end one does not obtain genuine time travel in Deutsch's account.
The systems in question travel from one time in one world to anoth-
er time in another world, but no system travels to an earlier time in
the same world. (This is so at least in the normal sense of the word
'world', the sense that one means when, for instance, one says 'there
was, and will be, only one Elvis Presley in this world'.) Thus, even
if it were a reasonable view, it is not quite as interesting as it may
have initially seemed.

Conclusions
What remains of the killing-your-earlier-self paradox in general rel-
ativistic time travel worlds is the fact that in some cases the states on
198
 Time Travel and Modern Physics

edgeless spacelike surfaces are 'overconstrained', so that one has

less than the usual freedom in specifying conditions on such a sur-
face, given the time-travel structure, and in some cases such states
are 'underconstrained', so that states on edgeless space-like sur-
faces do not determine what happens elsewhere in the way that
they usually do, given the time travel structure. There can also be
mixtures of those two types of cases. The extent to which states are
overconstrained and/or underconstrained in realistic models is as
yet unclear, though it would be very surprising if neither obtained.
The extant literature has primarily focused on the problem of
overconstraint, since that, often, either is regarded as a metaphysi-
cal obstacle to the possibility time travel, or as an epistemological
obstacle to the plausibility of time travel in our world. As we have
discussed, using responses and counterresponses, it is not entirely
clear that it is indeed an epistemological or a metaphysical obstacle.
It is true that our world would be quite different from the way we
normally think it is, if states were overconstrained given the time
travel structure. If anything, underconstraint seems even more
bizarre to us than overconstraint. However, time travel is quite
strange to begin with, and it does not appear to be a terribly strong
additional argument against time travel that it has strange conse-
quences.

Bibliography

Deutsch, D. 1991. 'Quantum mechanics near closed timelike curves,'

Physical Review D 44, 3197-217.
and Lockwood, M. 1994. 'The quantum physics of time travel',
Scientific American, March 1994, 68-74.
Earman, J. 1972. 'Implications of causal propagation outsider the null
cone,' in Foundations of Space-Time Theory, Minnesota Studies in the
Philosophy of Science, Vol VII, Earman, J., Glymour, C. and Stachel, J.
(eds), pp 94—108. Minneapolis, University of Minnesota Press.
Earman, J. 1995. Bangs, Crunches, Whimpers and Shrieks: Singularities and
Acausalities in Relativistic Spacetimes. New York: Oxford University
Press.
Earman, J. and Smeenk, C. 1999. 'Take a ride on a time machine,'
Manuscript.
Echeverria, E, Klinkhammer, G., and Thorne, K. 1991. 'Billiard ball in
wormhole spacetimes with closed timelike curves: classical theory,'
Physical Review D, Vol 44 No 4, 1077-99.
Friedman, J. et. al. 1990. 'Cauchy problem in spacetimes with closed time-
like lines,' Physical Review D 42, 1915-30.
Friedman, J. and Morris, M. 1991. 'The Cauchy problem for the scalar
199
Frank Arntzenius and Tim Maudlin

wave equation is well defined on a class of spacetimes with closed time-

like curves,' Physical Review letters 66, 401-4.
Geroch, R. and Horowitz, G. 1979. 'Global structures of spacetimes,' in
General Relativity, an Einstein Centenary Survey, Hawking, S. and
Israel, W., eds.
Godel, K. 1949. 'A remark about the relationship between relativity theory
and idealistic philosophy,' in Albert Einstein: Philosopher-Scientist, edited
by P. Schilpp, pp. 557-62. Open Court, La Salle.
Hocking, J., and Young, G. 1961. Topology. New York: Dover
Publications.
Horwich, P. 1987. 'Time travel,' in Asymmetries in time. Cambridge: MIT
Press.
Malament, D. 1985a. '"Time travel" in the Godel universe,' PSA 1984,
Vol 2, 91-100. Asquith, P. and Kitcher, P. editors. Philosophy of Science
Association, East Lansing, Michigan.
1985b. 'Minimal acceleration requirements for "time travel" in Godel
spacetime,' Journal of Mathematical Physics 26, 774—77.
Maudlin, T. 1990. 'Time Travel and topology,' PSA 1990, Vol 1, 303-15.
Philosophy of Science Association, East Lansing, Michigan.
Novikov, I. 1992. 'Time machine and self-consistent evolution in problems
with self- interaction,' Physical Review D 45, 1989-1994.
Thorne, K. 1994. Black Holes and Time Warps, Einstein's Outrageous
Legacy. W.W. Norton: London and New York.
Wheeler, J. and Feynman, R. 1949. 'Classical electrodynamics in terms of
direct interparticle action,' Reviews of Modern Physics 21, 425-34.
Yurtsever, U. 1990. 'Test fields on compact space-times,' Journal of
Mathematical Physics 31, 3064-78.

200
Freedom from the Inside Out
CARL HOEFER

0. Introduction

Since the death of strong reductionism, philosophers of science

have expanded the horizons of their understandings of the physical,
mental, and social worlds, and the complex relations among them.
To give one interesting example, John Dupre has endorsed a notion
of downward causation: 'higher-level' events causing events at a
'lower' ontological level. For example, my intention to type the
letter 't' causes the particular motions experienced by all the atoms
in my left forefinger as I type it. The proper explanation of the
motions of an atom at the tip of my forefinger primarily involves
my intentions, rather than (for example) the immediately preceding
motions of other nearby atoms, or any other such particle-level
events.
While this is a natural enough idea on the face of it, such down-
ward causation has seemed to be in tension, or outright conflict,
with another compelling intuition, which Dupre calls causal com-
pleteness.

'This is the assumption that for every event there is a complete

causal story to account for its occurrence. Obviously enough, this
is a view of causality the roots of which are to be found in the soil
of determinism. The paradigm of a complete causal story is the
sufficient (and perhaps even necessary) antecedent condition pro-
vided by a deterministic causal explanation. However ... [since
microphysics seems likely to be indeterministic], it is important
to consider theindeterministic analogue of deterministic causal
completeness. It is not hard to see what this should be. The basic
idea is that there should be some set of antecedent conditions that
together determine some precise probability of the event in ques-
tion.' (Dupre 1993, pp. 99-100)

As Dupre points out, belief in some such doctrine as causal com-

pleteness underlies the attraction many philosophers still feel for
reductionism, despite the latter's untenability in any form stronger
than mere supervenience.
It is causal completeness that is at the heart of the age-old
201
Carl Hoefer

dichotomy between free agency and physical determinism. For if

determinism is true, there is a prior, sufficient cause of my finger-
tip's atoms' motions: the earlier state of the physical world. 'Earlier'
could mean mere moments ago, or it could mean at some time
before I was born. Given this prior, sufficient cause, ray intentions
seem idle and epiphenomenal; they are there, to be sure, but they
are just as much caused by this prior physical state of affairs; my
'free' will seems then a hollow joke.
But determinism is not necessary for the threat of causal com-
pleteness to free agency to arise. For reasons that Kant first realized,
indeterminism at the microphysical level does not seem to help. The
randomness, if any, in microscopic phenomena does not seem to
'make room' for free will, but rather only replaces a sufficient phys-
ical cause with (at least in part) blind chance. The presumption in
favour of upward causation and explanation (from microphysical to
macrophysical) that comes with causal completeness is what cuts
free agency out of the picture, whether this causation is determinis-
tic or partly random.
Philosophical subtlety has thus put our freedom in double
jeopardy.' Some philosophers of a pluralist/empiricist bent
(Cartwright and Dupre for example) respond by saying that they
trust the evidence of common sense more than such philosophical
subtleties. Maybe there are really no laws of nature at all, in the
strict sense; maybe causation at the micro-physical level deserves
no priority over causation in the form of human agency. At any
rate, on any viable concept of evidence, they say, the evidence in
favor of free will is stronger than the evidence for universally true
physical laws.2
On this last point, I think they are wrong: we do have very strong
evidence for universal, exceptionless laws of nature. But fortunate-
ly, skepticism about true, universal laws of nature is not necessary
to derail the apparent challenge to free agency coming from causal
completeness. All that is needed is a proper understanding of time—
what it is in the physical world, what it is in human affairs, and how
they are related. Given the proper understanding of time, we will
1
There is of course a long tradition of philosophers responding to this
apparent threat by arguing for the compatability of freedom and deter-
minism, when the former is properly understood. For what follows, I need
not enter into these debates. See Fischer (1994) for a comprehensive and
robust defense of (a form of) compatibilism.
2
Dupre does in fact say exactly this, in his paper 1996. Cartwright has
not discussed free will explicitly, to my knowledge, but her views about
laws, causation and evidence seem to fit well with this response.
202
 Freedom from the Inside Out

see that freedom and determinism are compatible—compatible in a

much more robust sense than has ever been thought possible.3

1. The Two Times

'Time' means one thing in physics, and something quite different in

everyday human affairs. McTaggart first described the distinction
clearly, and gave the two times names: A-series time and B-series
time.
A-series time is the time in which we live our lives. There is the
past, the present (the 'now'), and the future; and the present 'moves'
inexorably into the future leaving more and more of our lives
behind us. B-series time is by contrast 'static': time is a linear order-
ing (or partial ordering), typically represented by a line on which
each point represents an instant of time, but no point is distin-
guished as 'now'. (See figure 1.) Things may change in B-series
time, by having one set of properties at one point, and a different
set of properties at a later point. But time itself does not 'change' or
'move'. Physics seems to describe the world entirely in B-series
terms, and to have no need of A-series concepts such as present and
future. Indeed, many philosophers believe that physics since
Einstein's 1905 relativity theory is outright incompatible with the A-

A-series time
Past Now Future

B-series time
1900 1950 2000 2050 2100 2150
Figure 1. The two times.

3
On a topic such as freedom of the will, it is too much to hope that any
proposal can be completely novel. In section 4 I link my proposal to Kant's
reconciliation of freedom and determinism. Further, as I recently discov-
ered, Peter Forrest (1985) has defended free will along lines similar in
many respects to those developed here. There are however quite substan-
tial differences, and in particular Forrest does not bring considerations
about time into play in his account.
203
Carl Hoefer

series.4 Regardless of whether this is correct or not, it is still true

that physics does not require A-series time notions, and seems to
find a natural fit only with B-series time. (A possible exception to
this is in quantum mechanics, but only under the most bizarre
(many-world) or idealistic (consciousness-collapse) interpretations.)
When space is combined with B-series time explicitly, as
Minkowski first did in 1908, we get a description of the world as a
whole, with four dimensions. This is what we do in drawing
'Minkowski diagrams' in relativity theory, but it works equally well
from the perspective of Newtonian physics. Either way, philoso-
phers have found it useful to think of the world, consisting of 3 spa-
tial dimensions and one (B-series) temporal dimension, as a 'block
universe' (figure 2).
In the block universe, time is certainly to be singled out as differ-
ent from the three spatial dimensions—in terms of the laws (if any),
the metrical structure(s) giving spatial and temporal distances
between events or world-points, and so on. Likewise, at least over
the part of the block accessible to our observations, the two direc-

i Increasing
time

X (one spatial dimension)

Figure 2. The block universe.

4
The reason is that the A-series seems to require a privileged way of
dividing events into those happening now, vs. those in the past or future,
which is effectively a privileged notion of absolute simultaneity. Special
relativity, as standardly interpreted, is incompatible with an absolute stan-
dard of simultaneity.
204
 Freedom from the Inside Out

tions of time (past-directed / future-directed) are distinguishable.

But what is not to be found is an ontological separation of parts of
the block into past, present and future. This striking fact means that
events of 1000 years in Earth's future are, in terms of reality or exis-
tence, no different from the events {now) of your reading these
words, or the events of last week.
This notion is hard to grasp, and feels threatening to us as free
agents. It has even been advanced, incorrectly, as a vindication of
fatalism. For, viewing ourselves and our actions from within the A-
series perspective, we think of future events as open in some real
sense, to be determined (partly) by our choices. But in the block, all
events are equally real, those in your far future no different from
those in your past.5 This 4-D block world that physics offers us
seems impossible to reconcile with this agent-centred, A-series-
embedded perspective.
I will now argue that in fact, matters are just the opposite of how
they seem. The very 'timelessness' of the 4-D block (in an A-series
sense) leaves us free to reject the customary view that past events
determine present choices. From the B-series perspective there is no
reason to think of past -^ future determination as more important
or real than future -^ past determination. And, even more to the
point, one can equally view a set of events in the middle as deter-
miners of both past and future events.
This is exactly what we should do. Our free actions, intentions,
thoughts etc., in the middle of the block universe, are part of what
determines how the rest of the block shall be. In order to make the
point as clearly as possible, I will first discuss things under the
assumption of some standard, Newtonian-style determinism. The
idea here is that given the complete state of affairs 'at a time' in the
universe (i.e., all physical facts specified on a time slice or thin sand-
wich), plus the true laws of nature, all earlier and later physical
events are logically determined. 6 Weaker forms of determinism can
5
See Horwich (1987) Asymmetries in Time for the correct refutation of
the argument for fatalism ('logical' fatalism) based on the block universe.
6
See Lewis (1994) for an explication of determinism in terms of possi-
ble worlds, and Earman (1986) for detailed discussion of the difficulties of
defining and assessing determinism in various physical theories. The
strong 'Newtonian-style' determinism I am assuming for the moment
turns out, as Earman shows, to be best captured not in Newtonian physics
but rather in Special-relativitstic physics. For the discussion to follow, I am
assuming bi-directionality of determinism. This is assured by time-rever-
sal invariant physical laws, but this condition is not needed. For example,
Callender (2000) argues that QM is not time-reversal invariant. Under a
Bohmian interpretation, it is nevertheless bi-directionally deterministic.
205
Carl Hoefer

be defined, but they pose, prima facie, less of a problem for free
agency.7 Later I will come back to freedom in a causally complete
but probabilistic world.

2. Freedom from the Inside Out

Determinism tells you that the state of the world at a time deter-
mines all the rest, past and future, but it doesn't tell you which slice,
if any, explains or determines all the rest. The challenge to free will
from determinism has not come from the physics, but rather from
the unholy marriage of deterministic physics with our A-series view
of time. The worry we have is that a past slice (long in the past,
maybe even the 'initial conditions' of the universe if there are such)
determines our actions now. We never think of a now-slice (includ-
ing the voluntary actions we perform now) determining what hap-
pened in the past. Why not? There are two reasons. First, we uncon-
sciously assume a metaphysical picture that is A-series based and
incompatible with the block universe: we think of the past as 'real',
fixed or determinate, the present as also 'real' (or becoming so), but
the future as 'indeterminate' or 'open'. And as the zipper of the now
moves into the future, it's the future that is getting determined, not

Figure 3.
7
See section 2.1 below for further discussion of this point.
206
 Freedom from the Inside Out

the past. Once one unearths this lurking metaphysical picture, its
irrelevance becomes obvious. Physics has no truck with any of it,
and (as noted before) is probably incompatible with it, when under-
stood as applying to physical events per se. From the B-series or
block-universe perspective, there is no reason to think of the past as
determining the present and future, rather than vice-versa, and so
on.
The second reason is more interesting. When we consider the
idea of events in a time slice now physically determining the past,
we become nervous because it looks as though we are positing back-
ward causation. So if I assert that my actions, now, are free and
explained or determined only by my own will; and that in a deter-
ministic world, that may entail consequences about the past, but so
what?; it looks as though I am positing backward causation, and giv-
ing myself the power to affect the past. And this is thought to be
unacceptable on solid physical grounds, independent of any A-
series/B-series considerations. In many presentations of the incom-
patibility of determinism and free will, this worry about affecting
the past comes out explicitly: saying 'I could have done otherwise'
is analysed as tantamount to saying 'I could either have caused a law
of nature to be violated, or changed the past.' 8 Laying to rest this
worry that freedom with determinism must involve either unac-
ceptable backward causation or 'changing the past' will be the first
task of the next section.
The idea of freedom from the inside out is this: we are perfectly
justified in viewing our own actions not as determined by the past,
nor as determined by the future, but rather as simply determined (to
the extent that this word sensibly applies) by ourselves, by our own
wills. In other words, they need not be viewed as caused or explained
by the physical states of other, vast regions of the block universe.
Instead, we can view our own actions, qua physical events, as pri-
mary explainers, determining—in a very partial way—physical
events outside ourselves to the past and future of our actions, in the
block. We adopt the perspective that the determination or explana-
tion that matters is from the inside (of the block universe, where we
live) outward, rather than from the outside (e.g. the state of things on
a time slice 1 billion years ago) in. And we adopt the perspective of
downward causation, thinking of our choices and intentions as pri-
mary explainers of our physical actions, rather than letting
microstates of the world usurp this role. We are free to adopt these
8
For example, see van Inwagen (1975). For a thorough and illuminating
treatment of the challenge to free will from determinism, see Fischer
(1998).
207
Carl Hoefer

perspectives because, quite simply, physics—including our postulated,

perfected deterministic physics—is perfectly compatible with them.
As I said before, exploring the consequences of these perspec-
tives and defending them against apparent problems will occupy
most of the rest of the paper. But the key to the defense has already
been explained, and needs repeating. The notion of past events
determining and explaining future events, and the opposite direction
(or an 'inside-out' direction) of explanation being somehow wrong
or suspect, arises completely from an unholy marriage of A-series
time with deterministic physics. The mistake is natural and under-
standable, because of the way the A-series dominates our lives and
our thinking, especially causal/explanatory thinking. It remains
nevertheless a mistake. A deterministic physics gives us logical rela-
tions of determination, not a unique temporal relation of determi-
nation. In the block universe one can view a slice now, or a future
slice, or a future '/2-block, or the past '/2-block 'before' now, as logi-
cally determining the rest.9 These logical relations however are not
in any interesting sense explanatory, nor even causal. Physics does
not pick any one out as more important than the others, and indeed,
equally allows us to ignore all of them when it comes to thinking
about causation and explanation in things that matter in our lives,
leaving room for downwards causation, as we will see below.
This is not the way we are accustomed to thinking of determin-
ism. We usually stay in our A-series perspective on the world, tacitly
conflate determination with causal explanation10, and there we are,
mired in the apparent incompatibility of determinism with our
actions' being explained by our choices. A first antidote to this mis-
take is firmly to keep in mind that physical determinism belongs in
the B-series world of physics alone. To break the conflation between
determination and causal explanation, it helps to remember that
deterministic physics equally allows future -^ past determination, but
it does not thereby tell us that the future causally explains the past.
The full antidote can only come by exploring the consequences of an
'inside-out' perspective on determination, and making sure that they
are acceptable both physically and for common sense.

2.1 Past -> future determinism only? Above I said that weaker forms
of determinism than the full time-symmetric Newtonian type we
have been assuming pose, prima facie, less of a threat to free will.
But do they really? In particular, does my argument for freedom
9
10
Always assuming the truth of the deterministic laws, of course.
Often the conflation is explicit, as the phrase "causal determinism"
indicates.
208
 Freedom from the Inside Out

from the inside out still have plausibility, if what physics gives us is
past -> future determinism but not future (or middle) -^ past deter-
minism? At first blush, it might seem that the plausibility of the
perspective on offer is undermined. A closer look remedies this mis-
apprehension.
'Past -> future determinism only' means that the future -^ past
relationships allowed by the laws are one-many, while the past -^
future relationship is one-one. These relationships are still, however,
logical rather than causal or explanatory. As long as our physics
remains fully expressable in terms of B-series time, and has no need
of A-series time, the one-way character of their determinism does
not mean that the past is 'fixed' in some sense vis a vis future events.
Nor do past events become somehow 'logically prior' to present and
future events in the block. It is true that we can say that the past
(plus the laws) entails our present actions, and can not any longer
make the same claim regarding the future (which claim, psycholog-
ically, perhaps helps break the grip of the idea that these determi-
nation relationships render us unfree.) But this change does nothing
to weaken the claim that the physical world's time is B-series time,
in which past, present and future events all have the same ontolog-
ical status. It does nothing to re-assert the notion that the past is
'fixed, done, and beyond our control'. In short, because this
hypothesized weaker form of determinism does not re-impose an A-
series metaphysics of time on us, it does not at all undermine the
perspective of freedom from the inside out.
In fact, in terms of the worries for this perspective that we are
about to explore, past -> future only determinism reduces their
strength. We are about to consider worries that arise if we consider
our free actions as prime, explanatory starting-points, having con-
sequences toward both past and future. But as just noted, under
past -> future only determinism, the present—past relationship is
one-many rather than one-one. So whatever the constraints our free
actions place on the past turn out to be, in principle they will be
weaker than they would under full, bi-directional determinism. The
comment made at the end of §1 was correct: the challenge to free
will posed by weaker forms of determinism is in fact weaker.

3. Causation and Consequences

Can this 'inside-out' perspective be held, though? Does it not make

the mistake of claiming that our actions now have causal conse-
quences toward the past? The answer is 'No'. From the inside-out
209
Carl Hoefer

perspective, our freely chosen actions place constraints on what the

past and future can be like, but the constraints are astonishingly
weak, both toward the future and (especially) toward the past.
To discuss the question in more detail, let's assume that a human
action (including the perceived surroundings of the agent's context)
is a physical event type that has innumerable instantiations at the
microphysical level. We assume, in other words, that there is some
ill-defined and probably infinite set of microphysical state-types that
are 'good enough' to count as a supervenience base for my typing 't'
in the assumed context. In doing so, we are doubtless assuming a
more reductionist picture than is likely to be true, but this is needed
in order to make the apparent challenge as strong as possible."

3.1 Consequences toward the past

If I freely choose to type this letter, Y, the choice in its context

entails that some one of this enormous number of micro state-types
shall be, and that is all. The constraints this places on how the past
should be, even (say) the past of only one minute earlier, are prob-
ably either trivial or non-existent. Thinking of the constraints
toward the future helps illustrate the weakness in either direction.
In his famous 1908 paper on causation, Russell pointed out that
no cause ever guarantees the following of the customary effect -
unless we inflate what we count as the cause to make it identical to
a time-slice of the whole state of the world over a huge region of
space. I reach for the 't' key, I depress it; will a 't' appear on the
screen microseconds later? Typically, of course, yes; but not as a
matter of logic (plus the laws), unless we rule out all possible inter-
ferences that could intervene and prevent the effect. (Think of
every way that the computer might malfunction, or the power be
cut, or a black hole whiz through your CPU at exactly the right
time, ...)
The same goes toward the past: in terms of logical determination,
our actions have little or no necessary consequences about what the
past shall be like, outside of what is already presupposed in describ-

11
Here I am giving the benefit of the doubt to supervenience on the
microphysical, and interpreting it relatively strongly. The idea is that,
although there can be no conceptual reduction of 'Carl types the letter 't'
on his laptop' to the language of microphysics, it is nevertheless the case
that at least God could say, if you showed him a microstate, whether or not
it is good enough to count as 'Carl typing the letter 't' on his laptop' (or Ct
for short), in the context.
210
 Freedom from the Inside Out

ing the context. At the microphysical level the constraint is just that
earlier microphysical states have to be logically consistent with a
microstate of the correct type (i.e., one corresponding to my typing
a 't') obtaining, at the time and place that it does. If the microstates
we are positing cover, for example, a spatial area of 10 metres radius,
then any given microstate logically entails the earlier microstates (i.e.,
toward the past) over an ever-shrinking spatial region, which vanish-
es 'after' a time period exactly equal to the time that light takes to
travel 10 metres. (See diagram 4.) Specifying the microstate over a
region of space and a slice or sandwich of time, in other words, logi-
cally determines the past and future microstates only over symmetric
past- and future-pointing 'light cones' which exist only for an absurd-
ly short period of time. All this is so, assuming Special Relativity's
restriction on the velocity of physical things. If we remove that
restriction, then the regions of past and future logical determination
vanish entirely. And when we recall the huge (probably infinite) num-
ber of microstates that can serve as basis for a macroscopic event (my
typing the 't'), the logical determination toward the past is corre-
spondingly decreased. (When I freely choose to type the 't', I do not
thereby choose to actualize a particular microstate!)
Despite the correctness of all this in logical/physical terms, we

S\ +t -33.3 * 10A-9 sec

time / \

\ (10 metre radius)

Spatial \
v
dimensions >

—t -33.3 * 10A-9 sec

Figure 4. Space-time regions determined by events on hypersurface of

10m radius.
211
Carl Hoefer

nevertheless have to acknowledge that the region of 'practical deter-

mination' of the state of things, toward the future, is usually much
greater. There are pervasive and fortunate circumstances in the
physical world that allow it to be the case that interferences such as
Russell noted are rare, and that we are usually successful in pro-
ducing the effects that we want toward the future. After all, usually
my computer is functioning perfectly, there are no black holes or
meteors or laser bolts heading toward me, etc. So usually, when I
type 't', a 't' appears and stays there for a while. We are able, fortu-
nately, to make things be the way we wish at the macro-level, more
or less completely - depending on what we're aiming for—and for a
goodly amount of time. If the same thing were true toward the past,
then if freedom from the inside out were the case, we should have
the ability to freely choose to make past events be the way we wish
(most of the time, to some limited extent). This would quickly lead
to paradoxes of the time-travel variety. For example, having
observed the word 'example' on my screen for the past minute, I
could (it seems) now take some action that causes the screen to be
blank for the past minute. This means either postulating a 'chang-
ing of the past'—which is incoherent, or at the very least takes us
outside of the block-universe perspective we have been assuming -
or mysterious interventions that prevent us from succeeding in our
backward-effect actions. (Banana peels are the standard mechanism,
in the time travel literature). But fortunately, the same thing
(ability to cause large-scale, enduring effects most of the time) is not
true toward the past.
Temporally asymmetric features of our world make it very
unlikely that our free actions leave 'traces' on the past of a macro-
level and repeatable nature. Philosophers interested in the 'direction
of time' problem have documented some of these circumstances in
depth: the thermodynamic asymmetry, the 'fork' asymmetry, the
knowledge asymmetry, the radiative asymmetry, and so on. Here is
an example. We know that if we want to have a drink be at a uni-
form 2°C, we can start with our mixture at room temperature, add
lots of ice, and wait. We don't have to worry about the drink getting
hotter and the ice bigger. But suppose that on alternate days, the
Second Law of thermodynamics switched temporal directions.
Then on alternate days, we could cause a nice cool drink to have
been present earlier, by adding ice to a room-temperature mixture.12

12
Here I am glossing over all the thorny problems about whether human
bodies could live under such a reversal of thermodynamic asymmetry, and
whether the perceived flow of time would not then reverse as well.
212
 Freedom from the Inside Out

But things being how they are, thermodynamically, we can't do any-

thing of the sort. I can add ice to the mix, but nothing at all is then
entailed about the past features of the drink—even assuming the
absence of external influences. It might have sat there for a day, at
equilibrium; or it might have had ice in it two hours ago; or it might
have been quite hot; and so on.
It appears that our inability to produce causal effects toward the
past is largely due to (1) these pervasive asymmetries in physical
phenomena; and (2) the nature of our conscious experience and of
sensation, which are either 'in', or somehow produce, the A-series,
flowing time of common sense.131 regard these as very puzzling and
unresolved issues; fortunately they do not need to be resolved for
our purposes here. What matters is that our free actions, while they
may have logical consequences about the past because of determin-
ism (of a highly disjunctive nature, and for a trivial amount of
time), do not have to be thought of as causally bringing about large-
scale features of the past, or as explaining them.
Finally, notice that if all this was incorrect, the traditional picture
would be in trouble also. Suppose we decided that, on the assump-
tion of freedom from the inside out, we would in fact be able to
effect noticeable backward causation. In that case, merely rejecting
the perspective of freedom from the inside out would not automat-
ically make this backward causation go away! Sticking with the same
deterministic physics, the physically possible worlds with the back-
ward-causation events would still be physically possible even if we
stuck to thinking of determinism as a past —> future relation. The
human actions producing backwards effects would still be physically
possible. So if backward causation is a worry for one perspective, it
should be a worry for both. Further constraints on physically pos-
sible worlds would have to be added to eliminate the threat, and
their justification would not be from physics alone. But let's leave
this concern for now, and carry on assuming that backward causa-
tion is not a worry.

3.2 Harmony
Granting this, one still might have some worries about harmonisa-
tion—about whether all the different actions we believe ourselves to
be able to freely choose, can really fit together (a) with each other
and (b) with the past as we know it (i.e., the macroscopically

13
See Horwich (1987) and Price (1996) for extended discussion of tem-
poral asymmetries.
213
Carl Hoefer

described, known past) under determinism. One wonders whether

billions of humans, all exercising free will, over the course of mil-
lenia, shouldn't be expected to generate enough consequences
toward the past to generate contradictions—despite the weakness
and disjunctive nature of the consequences of each act taken on its
own. My freely chosen actions don't just have to harmonise with
my immediate past; they have to harmonise with your immediate
past and everyone else's, and they all have to be able to be fit
together into a consistent past history of the world. The worry
here is this: how do we know that there is always at least one
microstate of the whole past that is compatible with the conse-
quences (toward the past) of all the freely willed actions of all
agents in history? Might it not be, instead, that once the free
choices of (say) 4 billion humans are conjoined, then the possible
choices of the rest of humanity are either removed (only one over-
all microstate is compatible) or severely constrained (each of us
has few genuine choices available to us)? Given that everything has
to fit together in such a way as to not violate the physical laws, one
may worry that there needs to be a pre-established harmony (or,
better: a harmony simpliciter—the 'pre' is misleading), and that
because of this, we really are not free to do all sorts of things after
all.
We know, of course, that all actual choices in fact fit together har-
moniously; this is our starting assumption, that a deterministic
microphysics holds sway over all actual events. So this harmony
worry really has to do not just with actual choices, but the alterna-
tive choices we think we could have made: our freedom to do other-
wise. This then brings us to the heart of the issue of the compati-
bility of freedom and determinism: the counterfactuals we believe,
the could-have-done-otherwise's.
When I type the letter 's' I may think that I could have chosen
to type a 'z' instead, in keeping with my nationality. And I think I
could have done so, with the past being, macroscopically, just the way
I know it to be. But can I really? Or is it instead the case (though
we can't of course see why) that for me to type that 'z' instead, the
past would have to have been different macroscopically (e.g., I
would have had to have had corn flakes for breakfast instead of
toast)?
The qualifier 'macroscopically' is absolutely crucial here. For
note that although we are sticking to a block-universe perspective
when it comes to real physics, and hence not supposing that the
actual physical state of the world this past morning is somehow
ontologically privileged over present or future states, nevertheless in
214
 Freedom from the Inside Out

terms of our actions as we conceive them, there is an important asym-

metry. We think of ourselves as beings with a certain history, in a
physical world with its own history, and our actions as arising freely
given (or despite) all that.'4 And if this perspective was not in fact
sustainable, then the compatibility of freedom with determinism I
am after would not be possible after all.15 I think I have freedom of
the following kind: even given that the past history of the world is,
macroscopically, as I (and indeed every other agent) knows it to be,
I can either type the V or the 'z' (depending on which I choose).
Can the past and our present actions, as well as those we don't choose
but think we could, all fit together harmoniously in the way this con-
ception of freedom demands? Can everything harmonize as well as
harmonise?
Part of the response to this worry is what has already been
explained: that logically each person's free actions entail only (at
most) that one of an enormous set of past microstates obtain, and
that only over a time-span that is vanishingly small. The time-
asymmetry of typical physical events further rules out that there
should be macro-scale consequences toward the past under 'typical'
circumstances. If each person's free actions entails practically noth-
ing about the past, it is plausible that all persons' actions conjoined
should be able to fit together consistently. Moreover, of course,
looking at the actual world from the block-universe perspective, all
human actions do fit together consistently. So we have one example
of a universe where it all works. The worry of course is that only one
such, or very few such worlds are possible given the laws and the
contextual/historical circumstances of our free choices. But what
reason can we have for this worry?

14
Here I am implicitly offering a criticism of one standard way (Lewis')
of analysing counterfactual statements. Lewis, who seems inexplicably
wedded to the A-series in all his metaphysics, supposes that in most uses
of counterfactuals we mean to hold the past fixed—and I agree. But for
Lewis this means the physical past in all its gory microphysical detail; so if
determinism is true, it takes a miracle to get the if-had-done-otherwise
scenario started. But why hold the past fixed in microphysical detail? What
matters for action is the macroscopic past, that we know about empirically.
When only that is fixed, I suggest, we don't need miracles to postulate
various different actions and their likely future consequences.
15
In this case, we would have to live with the threat to freedom posed by
causal completeness, or take up a different compatibilist picture, such as
that of Fischer (1994). He argues that the freedom-relevant sense of con-
trol over one's actions is 'guidance control', which does not require the
ability to have done otherwise.
215
Carl Hoefer

If anything, it seems that evidence points strongly the other way.

I can test my free will right now, in the very typical circumstance of
a person typing on a computer in a small room. I type various
letters, randomly. Think of each letter struck as a run of an experi-
ment. The experiment is simply to see whether all sorts of letter-
producing choices, in a very normal physical context, starting from
macroscopically near-identical initial conditions, can fit together
consistently into one history. And the result is clear: they can.16
Extending this idea further, we can regard much of what happens in
an everyday life as providing similar evidence for harmony between
a given, fixed macroscopically-described past and multiple present
choices. I go to the 4th-floor cafeteria every day, and the menu on
offer is always the same; but my choices vary.
Someone gripped by the harmony worry here will say that all
this shows nothing. For each letter typed and each lunch selected
may not be in fact freely chosen, but rather determined by the
requirement of there being a globally consistent history (even
when only some of the past, macroscopically described, is held as
fixed). I find this worry very implausible, verging on the paranoid.
The idea is that somehow, the deterministic physics we are assum-
ing allows a world that is (toward the past) macroscopically like
ours, in which I type 't' here, but does not allow one in which I
type 'q' in that same place. Remember, we are not concerned with
the actual past history of the world in all its microscopic detail;
that does, of course, determine the present including that typing of
't'. We have set aside this traditional problem by adopting the per-
spective of freedom from the inside out. Instead we are here only
concerned with whether there should be a physically possible
world similar to actuality in some gross, macroscopic ways, and in
which I (or my counterpart, if you like) types 'q'. How could it be
the case that physics makes room for the one, but not for the
other?
The harmony worry thus boils down to this: that our posited
deterministic physics may allow vastly fewer possible worlds than
we can imagine, so few that our normal conception of the could-
have-done-otherwise is mistaken. And so few that what seems like
good evidence for multiple choices in a given context (such as the
evidence described above) is in fact not good evidence: there are

16
The point in this thought experiment is to keep the macroscopically-
described initial conditions as much identical as possible: not only the
room, lighting, etc. are the same, but also my intention—namely, to type a
letter at random.
216
 Freedom from the Inside Out

very few physically possible worlds like ours, even though at least
one of them (i.e., ours) happens to contain an abundance of mis-
leading evidence in favour of freedom. Without having a genuinely
adequate deterministic physics in hand to examine, there may be no
way fully to resolve this doubt. There might be no way even if we
did have the true physics in hand. But I am moved by the intuition
that in any recognizable deterministic microphysics, there will be so
many different micro-level world histories, there has to be more
than enough scope for freedom as we normally conceive it.

3.3 Indeterministic microphysics

When we turn to considering freedom from the inside out under the
assumption that an indeterministic microphysics holds in our
world, things become simpler in one sense, and more complicated
in others. Intuitively we expect the apparent challenge to freedom
posed by such an underlying physics—always less clear-cut than the
challenge from determinism—to dissolve more easily. Nevertheless,
care is required in thinking through the possibilities under indeter-
minism.
Again we insist on downward causation, and the need for past his-
tory (at the micro-level) to conform to the constraints set by the free
choices of agents. Again we suppose that there is a past micro-state
compatible with my typing 't' now, but also a macroscopically iden-
tical micro-state (which may or may not be different!) compatible
with my typing 'q' instead. But now the constraint is only that these
micro-histories must be consistent with our merely probabilistic
laws. Surely this is a looser set of constraints, and hence an easier
context in which to maintain freedom?
Perhaps, but this does not follow immediately and trivially from
the mere idea of an indeterministic microphysics. First of all, notice
that in a formal sense determinism could fail (and indeterminism
reign) without the challenge to free action changing significantly.
Suppose, for example, that it remains the case that the state of the
world (over the relevant region) a million years ago makes each and
every one of our actions have a probability greater than 99999%. I
submit that this does not alter the force of the traditional incom-
patibilist argument that we are unfree very much. But things could
be worse still; it might be that in fact our actions are all 100% neces-
sitated by the past of 1 million years ago. Suppose that indetermin-
ism holds only in this weak sense: once every 3 million years, an
atom of hydrogen pops into existence at a random location in the
universe; and this last happened 2.5 million years ago. Otherwise,
217
Carl Hoefer

events follow iron deterministic laws. This scenario posits what is,
in some formal sense, an indeterministic world; but in essence
things are just the same as under 'pure' determinism.17
But recall that we are advocating freedom from the inside out. It
is not necessary to maintain that we are only loosely constrained by
the past, to maintain that we have freedom. Instead we simply
maintain that the constraint goes the other way around. The past is
(partly) constrained by our choices. How will this partial constraint
play out under an indeterministic microphysics? Again, as above, it
is not possible to make definitive pronouncements without having
the physical laws before us (and it might be practically impossible
even then). Plausibility considerations are the best we can do.
Prima facie it seems plausible that an indeterministic micro-
physics, which allows (by definition) multiple futures branching
from a single past, should allow greater room for freedom than a
deterministic microphysics. We intuitively picture a 'branching
tree' structure of possibilities, and think of the forks as correspond-
ing to our free choices. The scenarios sketched above show us that
we cannot automatically assume this is so. What matters, then, are
the following questions: does our indeterministic microphysics
allow various worlds corresponding to a variety of free actions we
can undertake (in a given context) that all share an identical or
macroscopically identical past? And does it do so for all of our free
actions together, so that they harmonize appropriately?
What is needed, then, is the same as in the case of determinism:
a rich variety of physically possible worlds, so that we can take the
actual history as one among many similar possible histories, whose
actuality is explained (in part) by all our free choices. We need to be
able to say that, generally, we could have done otherwise in the cir-
cumstances where we normally believe this; this might or might not
imply that the past would have had to be different at the micro-
level. The 'might or might not' in the previous sentence is what dis-
tinguishes an indeterministic microphysics from a deterministic set
of laws. Those who equate indeterminism with automatic room for
freedom are assuming that these four words can be replaced with
'would not'. But this cannot be taken for granted.
What can be taken for granted is just the set of considerations
developed above in section 3.2. An indeterministic microphysics
might have a richer variety of worlds than a deterministic micro-
17
These brief remarks are meant to counteract a common tendency in
the free will literature, that of conflating indeterminism with a sort of
'anything goes' conclusion about what actions are physically possible given
a fixed (micro- and macro-) past.
218
 Freedom from the Inside Out

physics; but for all we know, it might not. To be the correct laws for
our world, it must allow a great deal of variety—including the
phenomena adduced in §3.2 as evidence that we should not worry
about harmony problems. In the end, then, the situation seems to be
the same as in the deterministic case.

4 Clarification of an Old Idea

Let me recap the main features of the notion of freedom from the
inside out. We carefully distinguish the true story of the physical
world as it is in itself, which is that of a block universe with only B-
series time, from the world of everyday experience and action, which
is wholly within A-series time. Physical determinism, if true at all,
is true of the block universe with its B-series time, and implies no
explanatory priority of the past over the future, or of future over
past, or of the middle over the far past and future. It is therefore
open to us to conceive of our actions as genuinely free, only properly
explained by our desires, beliefs and intentions despite being
logically determined by vast states of the world at other times.
While I have not seen this idea put forth in any modern discus-
sion of free will and determinism, I must confess that I believe the
first philosopher to advocate it was not me, but Kant. Kant
famously defended a metaphysical picture that postulated a
Newtonian/deterministic physical world, but also claimed that
rational beings were genuinely the authors of their own free actions.
How Kant thought he could reconcile these two theses is rarely dis-
cussed in a satisfactory way. A typical (and unsatisfactory) way of
reading Kant's suggestions on this point is to read him as claiming
that, in purely rational/intellectual terms, a person qua transcen-
dental being should be considered the author of his/her own
actions, at least for the purposes of praise and blame. But there are
also cryptic comments about the whole of a person's life actions
being but a single phenomenon, and a strong suggestion that the
non-temporality of the noumenal world is what allows us to think
of a person's will as the genuine source of their actions, despite
determination by past events in the phenomenal (A-series) world.
Here are some passages from the Critique of Practical Reason:
'... Repentance is entirely legitimate, because reason, when it is a
question of the law of our intelligible existence (the moral law),
acknowledges no temporal distinctions and only asks whether the
event belongs to me as my act, and then morally connects it with
the same feeling, whether the event occurs now or is long since
219
Carl Hoefer

past. For the sensuous life is but a single phenomenon in the view
of an intelligible consciousness of its existence (the consciousness
of freedom) ... [and] must be judged not according to natural
necessity which pertains to it as appearance but according to the
absolute spontaneity of freedom.' (LWB translation, p. 102)
'... [despite the determination of a person's actions], we could
nevertheless still assert that the man is free. For if we were capa-
ble of another view ... i.e., if we were capable of an intellectual
intuition of the same subject, we would then discover that the
entire chain of appearances, with reference to that which con-
cerns only the moral law, depends upon the spontaneity of the
subject as a thing-in-itself, for the determination of which no
physical explanation can be given.' (LWB translation, p. 103)
In other words, Kant suggests, the agent as a noumenal being
should be considered as the source and genuine explainer of his/her
own free actions, even though qua physical things in time their
actions are determined by earlier physical events.
Kant did not have McTaggart's distinction at his disposal. If we
bring it to bear on Kant's metaphysical picture, we can clarify (and
correct) that picture as follows. The block universe is the realm of
things in themselves, i.e., the world in itself, not as experienced by
consciousness. For Kant, 'time' meant A-series time, and that is
indeed restricted to conscious/rational experience. Physics, which
does try to describe the world in itself (contra Kant's epistemic
restrictions), needs only B-series time, i.e., a structure of relations
among events that underlies and is partly isomorphic to A-series
time. Rational agents can be understood as the ultimate explanatory
sources of their own free actions18; the rest of the noumenal world ,
i.e., the rest of the block universe, must simply be such as to accom-
modate those actions. The only real mistake Kant made was in the
locus of determinism: he thought it must be a feature of the world
of experience, due to the necessary conditions of possible experi-
ence. In fact determinism is no part of our experience of the world,
and if true at all, is only true at the subtle level of ultimate particles.
Nevertheless Kant seemed to have the fundamental point right:
18
At this point one perhaps wants to hear more about the positive char-
acterization of freedom that should accompany the negative picture (i.e.,
sketch of how the physical world leaves us room for freedom) developed
above. I will not try to sketch or defend any positive account, but I am
attracted to the basic idea we find in Kant: free action in the highest sense
is action that springs not from mere desire, but rather from something
intellectual, a concept of the good.
220
 Freedom from the Inside Out

when agents are conceived as 'things in themselves' (i.e., as rational

beings rather than as merely physical objects), their free actions are
quite compatible with overall physical determinism, because those
actions can be thought of as outside the time series (i.e., the A-series
with its allegedly fixed past) and hence not unfree despite being
'determined' by physical events lying to the past of them.
Whether or not this is really close to what Kant had in mind, I
think it is what we should believe to be the case, if our world is
causally complete. Free action and causal completeness are compat-
ible after all, and not in the (arguably) weak sense offered by tradi-
tional forms of compatibilism. You have choices, and you make
them. Because of determinism, your choices (like any events) place
constraints on what the world's history can be. But the direction of
determiniation (and, for most free actions, correct explanation) is
from your choices to the ways the physical world can be—both
toward the past and the future.
This picture of freedom from the inside out is more Idealistic
than some will find comfortable. Take a God's-eye perspective on
the block universe, and ask the question (Q): why are things as they
are in it? A 21st-century materialist is comfortable with this sort of
answer: 'Well, you see, there was this Big Bang at the beginning, and
after that things just sort of bump around in the ways permitted by
the laws of nature, and that leads to the whole history.' But this
answer (a) is infected with the A-series view of time, (b) seems to
presuppose an eliminativist picture of human thought and action,
and (c) begs the question "Why was the Big Bang just so and not
otherwise?" (a) is a mistake, (b) is at least dubious, and (c) is the
lump under the carpet which, if you try to flatten it, leads to moves
in all sorts of unpleasant directions (theology, Cosmic Anthropic
Principles, and so on).
I prefer the picture that starts with what we feel so strongly that
we really have: freedom to act in a variety of ways. This picture
places some constraints—probably only very weak ones - on what
an answer to this ultimate question (Q) can look like, if one is pos-
sible at all. It is Idealistic, in that the constraints involve giving
rational agents priority over trivia such as the physical micro-state
of vast regions of space-time (past or future). But this is a form of
Idealism that most of us can learn to live with. Appropriately, it is
McTaggart's distinction that helps us see that it is not nearly so
strange as it may at first appear.

221
Carl Hoefer

References
Dupre, John 1993. The Disorder of Things (Harvard University Press).
Dupre, John 1996. 'The Solution to the Problem of Free Will',
Philosophical Perspectives, 10, 385^02.
Fischer, John The Metaphysics of Free Will (Blackwell, 1994)
Forrest, Peter 1985. 'Backward Causation in Defence of Free Will',
MIND, 210-17.
Horwich, Paul Asymmetries in Time (MIT Press, 1987).
Kant, Immanuel. The Critique of Practical Reason, trans. Lewis White
Beck, Bobbs-Merrill Company, 1956.
Price, Huw Time's Arrow and Archimedes' Point (Oxford University Press,
1996).
Russell, Bertrand 'On the Notion of Cause', Address to the Aristotelian
Society, reprinted in Mysticism and Logic (Allen & Unwin, 1917).
van Inwagen, Peter, 'The Incompatibility of Free Will and Determinism',
Philosophical Studies 27, 185-99.

222
On Stages, Worms, and Relativity^

YURI BALASHOVt

Abstract
Four-dimensionalism, or perdurantism, the view that temporally
extended objects persist through time by having (spatio-)temporal
parts or stages, includes two varieties, the worm theory and the
stage theory. According to the worm theory, perduring objects are
four-dimensional wholes occupying determinate regions of space-
time and having temporal parts, or stages, each of them confined to
a particular time. T h e stage theorist, however, claims, not that
perduring objects have stages, but that the fundamental entities of
the perdurantist ontology are stages. I argue that considerations of
special relativity favor the worm theory over the stage theory.

1. Introduction
Recent work on persistence over time has produced a more fine-
grained inventory of views than we had a few years ago. Although
there are still two major rival accounts of persistence on the market:
three-dimensionalism (3D, endurantism) and four-dimensionalism
(4D, perdurantism), distinct varieties of each view have now been
identified. For example, philosophers who think that ordinary
material objects endure—that they are wholly present at all times at
which they exist—now explicitly include those who prefer to run
this position together with a certain theory of time, namely presen-
tism (roughly, the view that only the present exists),1 and those who

I am grateful to Michael Rea and Theodore Sider for valuable com-

ments on earlier drafts. An ancestor of this paper was read at the confer-
ence on 'Time, Reality, and Experience' (The London School of
Economics, September 2000). My thanks to the audience, and especially to
Simon Saunders and Steve Savitt, for very helpful critical discussions.
The work on this paper was supported by a junior faculty grant from the
University of Georgia Research Foundation.
T Department of Philosophy, 107 Peabody Hall, The University of
Georgia, Athens, GA 30605,USA. [email protected]
1
See, e.g., Merricks (1995, 1999), Hinchliff (1996), Zimmerman (1998),
Craig (2000), Markosian (forthcoming), and references to earlier work
therein. Prior's classic (1970) should be specifically mentioned.
223
Yuri Balashov

deny this link between the theory of persistence and the philoso-
phy of time. 2 Similarly, four-dimensionalists who think objects
perdure—persist by having different temporal parts at different
times—comprise those who think that this position presupposes
eternalism (the idea that all moments of time are on the same
ontological footing)3 as well as those who argue against this
connection. 4
Another important distinction, which has emerged within the
perdurantism camp, is between the worm theory and the stage
theory—the distinction with which I am concerned in this paper.
The worm theory features an ontology of 4D wholes (spatio-tem-
poral 'worms') occupying determinate regions of spacetime. Such
entitied have parts, or stages, each of them confined to a particu-
lar time (an instant or interval). The stage theorist, on the other
hand, claims, not that perduring objects have stages, but that the
fundamental entities of the perdurantist ontology are stages.5
In the next section, I attempt to draw the distinction between the
worm and the stage theories more precisely while explaining what
makes both of them varieties of four-dimensionalism. Then I offer
an argument defending worms over stages. The argument is based
on considerations of special relativity and requires, as a prerequi-
site, restating the stage view in the relativistic context. It will be
convenient to start with the classical framework, that of neo-
Newtonian spacetime, and then show what modifications are to be
made in the stage theory in effecting a transition to special
relativistic (Minkowski) spacetime.
Although I do not embark, in this paper, on the task of defending
four-dimensionalism as a whole against three-dimensionalism, I
believe that a somewhat similar argument could be applied to this
purpose. 6 My strategy presupposes the eternalist framework: it
2
See, e.g., Mellor (1981), Haslanger (1989), van Inwagen (1990), Rea
(1998).
3
Most four-dimensionalists are eternalists. Merricks (1995), who is a
three-dimensionalist, and Carter and Hestevold (1994), who maintain a
neutral position, have nonetheless argued for the link between perduran-
tism and eternalism.
4
Brogaard (2000) defends presentist four-dimensionalism. Not being
presentist four-dimensionalists themselves, Lombard (1999) and Sider
(2001, §3.4), have shown that this combination is consistent.
5
The stage theory has recently been elaborated and defended by Sider
(1996, 2000, 2001) and Hawley (2002). References to earlier works can be
found therein. The paradigm worm theorist is probably Heller (1990).
6
See, in this connection, Balashov (2000a, 2000b).
224
 On Stages, Worms, and Relativity

takes the 4D spacetime manifold of point events to be existing in

the fundamental tenseless sense and is not concerned to defend this
framework against presentism. One reason is that this paper takes
relativity seriously, and I agree with many writers7 that anyone who
takes relativity seriously cannot take presentism seriously. At the
same time, I believe, contrary to some,8 that taking relativity seri-
ously does not automatically force one into four-dimensionalism, let
alone a particular variety of it. One has to produce a substantive
argument to this effect.

2. Stages versus Worms

Both the stage and the worm theorists are four-dimensionalists

because both agree that temporally extended objects persist through
time by having temporal parts (stages) (cf. Sider 1996, 433). This
sets them against the paradigm three-dimensionalist who plainly

7
See, e.g., Savitt (2000), Callender (2000), Saunders (forthcoming), and
Sider (2001, §2.4).
8
Earlier work on persistence did not always draw a clear distinction
between four-dimensionalism, a particular ontology of material objects,
and eternalism or the 'block-universe' view, which is primarily a view
about time and the nature of events. See, e.g., Taylor (1955). Since special
relativity undoubtedly favours the latter, some writers were too quick to
conclude that it automatically favours the former as well. This link is
implicit in Quine (1950, 1987). Recent contributions to the persistence
debate have gone a long way towards dismantling this alleged package deal.
See, in this connection, Rea (1998), Balashov (1999, 2000a, and 2000c),
Sider (2001). Craig, a presentist endurantist, concedes that 'Embracing
spacetime realism does not... commit one automatically to ... four-dimen-
sionalism or perdurantism' (2001, 94n), but he contradicts himself else-
where: 'Spacetime realism [i.e., the view that all events populating the 4D
spacetime manifold tenselessly exist on the same ontological footing] raises
a host of problems due to its entailment of the doctrine of perdurance ..."
(2000, 124-5, my emphasis; cf. 2001, 192); 'If one is a spacetime realist,
then, barring conventionalism, things must have spatio-temporal parts'
(ibid., 202n68); 'A consistent spacetime realist will ... view objects as spa-
tio-temporal entities which perdure' (2001, 94n54). Craig's reasons for
thinking (in the end) that the combination of endurantism with eternalism
is inconsistent remain unclear to me, especially given that this combination
is widely accepted on the basis of a view of predication known as
Adverbialism (see Johnston 1987, Lowe 1988, Haslanger 1989, van
Inwagen 1990, Rea 1998). More on Adverbialism below.
225
Yuri Balashov

denies that the notion of temporal part makes good sense when
applied to objects (rather than, say, events).9
Furthermore, both the stage and the worm theorists typically
believe in temporal stages as well as 4D wholes. Indeed, a chief con-
temporary advocate of the stage theory, Theodore Sider, accepts
such wholes because they are aggregates of stages.10 And worm
theorists usually accept stages on the ground that the latter are parts
of what they take to be the central entities of their ontology. What
is, then, the difference between the two views?
As Sider notes (1996, 433), 'spacetime worms are [not] what we
typically call persons, name with proper names, quantify over'—and
attribute temporary properties to, one might add. Stages come
closer to filling this bill.11 To illustrate, consider the famous prob-
lem of temporary intrinsics. How can one and the same object—say,

9
By far, the strongest expression of this attitude belongs to van Inwagen
who has said of temporal parts: 'I simply do not understand what these
things are supposed to be, and I do not think this is my fault. I think no
one understands what they are supposed to be, though of course plenty of
philosophers think they do' (1981, 133).
10
'At one level, I accept the ontology of the worm view. I believe in
spacetime worms, since I believe in temporal parts and aggregates of
things I believe in' (Sider 1996, 433).
" Hawley (2002, Ch. 2) describes the difference between the worm and
stage views similarly:

According to [the worm] theory, persisting objects like bananas and ten-
nis balls are four-dimensional, and they satisfy certain predicates with
respect to certain times because of the properties of their temporal
parts. Alongside [the worm] theory, there is space for an alternative
account of persistence, one which retains the four-dimensional meta-
physics of perdurance theory whilst rejecting [the worm theory's]
claims about predication. ... According to [the] stage theory, nothing is
wholly present at more than one moment, so endurance theory is false.
But [the] stage theory also claims that the satisfiers of sortal predicates
like 'is a banana' and 'is a tennis ball' are momentary things, the very
things which instantiate ordinary properties like being yellow, being
spherical or being banana-shaped.
Consider the series of momentary stages whose sum is what [the
worm] theorists think of as the tennis ball. According to [the] stage
theory, when we talk about the tennis ball with respect to different
times, we talk about different stages in that series, and each of those
stages is a tennis ball. The tennis ball at one moment is spherical, and
the squashed tennis ball at another moment is not spherical: the spher-
ical tennis ball and the non-spherical tennis ball are different objects.
226
 On Stages, Worms, and Relativity

a poker—instantiate contrary properties at different times, such as

being hot at t\ and cold at t£ To solve this problem, the worm
theorist attributes hotness to the tx -stage of the poker worm and
coldness to its £2-stage. But it is the poker worm, not its stages, that
should properly be called the poker in the worm ontology. Clearly,
the 4D poker cannot instantiate temporary intrinsics simpliciter. It
can only do so derivatively, via its temporal parts, or stages. Thus it
is unable to do what the enduring poker can, according to the pre-
sentist three-dimensionalism, do quite naturally: by instantiating
hotness simpliciter at t\ and then (next morning) instantiating cold-
ness simpiciter at t2- Now this advantage is lost as soon as one
rejects presentism in favour of eternalism. In the eternalist setting,
the endurantist has to replace the simple properties hot and cold by
their time-indexed counterparts, hot-at-tj and cold-at-t2
(Indexicalism) or to replace the having of the usual properties sim-
pliciter by their temporally qualified having (Adverbialism).12 All
things being equal, it would be nice to avoid trading the intuition
about the having of temporary intrinsics simpliciter for the
(arguably unavoidable) rejection of presentism. And that is what the
stage view does. In saying that the poker is hot at 2j, the stage
theorist ascribes the usual (not time-indexed) property hotness to
the poker (i.e., the t] poker stage) simpliciter.
This example demonstrates the difference between the two vari-
eties of four-dimensionalism and suggests that, assuming eternal-
ism, the stage theory has an edge over both endurantism and the
worm view vis-a-vis the problem of temporary intrinsics. Sider fur-
ther argues that the stage theory offers the best unified solution to
the paradoxes of material constitution and coincident entities (1996;
2001, §5.8).
There is a price to be paid for these gains. To account for persis-
tence and change, the stage theorist is hard-pressed to adopt a tem-
poral version of the counterpart theory. The poker is hot tonight,
but it will be cold tomorrow morning. But the poker tonight is just
a poker stage confined to a certain time. How can it survive till
tomorrow and be cold then? Only by bearing a temporal counter-
part relation to another poker stage. The latter has a simple proper-
ty being cold, but it is the former, tonight stage (which, remember, is
the poker, on the stage theory) that has, vicariously, the temporal

12
For discussions of Indexicalism and Adverbialism, see Lewis (1986,
202-204, 1988), Johnston (1987), Lowe (1988), Haslanger (1989), van
Inwagen (1990), Merricks (1994), Rea (1998), Balashov (1999, 2000a,
2000c), Lombard (2000), Hawley (2002).
227
Yuri Balashov

property being cold tomorrow. The analogy to the modal counterpart

theory is obvious.
Those who are happy with the modal counterpart theory should
welcome the stage view as part of the package deal. Others, of
course, will consider this kind of commitment unattractive. But one
need not be a modal counterpart theorist to adopt the stage theory.
The temporal counterpart relation is a this-wordly affair, and real-
ism about other times and their denizens is not nearly as exotic as
full-blown realism about possibilia. Furthermore, transtemporal
aggregates of shorter-lived entities are familiar (e.g., from the per-
vading experience of temporally extended events, such as football
games and symphony concerts) in a way transworld aggregates of
possibilia are not.
In light of the claimed advantages of the stage view, it should be
given careful consideration, along with other views of persistence.
My eventual goal is to compare the stage and worm theories in the
relativistic context. This will require some preliminary work.

3. Local and Global Stages

The first thing to appreciate is that the notion of stage admits of a

global interpretation. More precisely, local object stages populate
and share global world stages.
According to the stage theory, various pieces of furniture in my
office are so many furniture stages. I can refer to them collectively
when I say, for example, that they overcrowd the room. What does
it mean, speaking stage-theoretically? It means that each of these
things (two chairs, three book cases, a desk, etc.) taken individual-
ly—each thing stage, that is—belongs to the room stage and shares
this larger stage with other things and with me (that is, my stage).
Similarly, the room stage belongs to the building stage, which itself
belongs to the campus stage (if there is such a thing). Everything on
the planet Earth populates the Earth stage, which inhabits the Solar
system, which, in turn, inhabits the Galaxy, and so on. All local
object stages, in short, share a global stage cutting across the entire
world. In fact, such a global stage is the world. Or, to borrow an
expression from Shakespeare and Sider (1996, 433), 'all the world's
a stage.'
Let us call objects (themselves stages, of course) populating and
sharing some global stage stage mates, by analogy with Lewis's
world mates. Consider an example (Figure 1). Descartes in 1620
228
 On Stages, Worms, and Relativity

(i.e., the 1620 stage of Descartes") and Galileo in 1620 are stage
mates because they share the 1620 global stage, which is the entire
world in 1620. They also share this temporal world with Kepler, but
not with Tycho Brahe. The reason is that the 1620 world stage con-
tains a Kepler stage, namely his 1620 stage, but it does not include
any stage of Brahe. This cosmologist does not exist in 1620. On the
other hand, Descartes in 1600 (i.e., his 1600 stage) and Galileo in
1600 share a global stage with both Kepler and Brahe (i.e., with
their corresponding stages). All four are stage mates; they coexist in
a single temporal world.

'(1596 -1650
1(1564 -1642

| Descartes
| Galiileo

1620

1600
L630)
^cho \ rahe
[546 601)

t .
i\
M-l i

in

Figure 1.

13
'The 1620 stage of Descartes' refers ambiguously to a year-long
Descartes' stage and to his momentary stage at a certain time in 1620.
Similarly with 'the 1620 world (or global) stage'. Hawley argues that stages
must be instantaneous: they 'need to be as fine-grained as possible change,
and thus they must be as fine-grained as instants, in order to account for
possible change in position' (2002, Ch. 2). Sider's works exploit the instan-
taneous sense too, and I will do the same in my analysis. I will also restrict
my consideration to point-like objects having no extension in space. The
reader will appreciate that nothing important turns on these idealizations.
But they do simplify the discussion significantly.
229
Yuri Balashov

The notions of existence and coexistence in a temporal world

require some comment.14 It may be objected that the only concept
of existence relevant to the 4D framework is the atemporal notion
that all spacetime entities, worms and stages alike, exist as parts of
the single reality, that of the 4D spacetime manifold. If that is the
case, then the distinction made above between Brahe's existence in
1600 and his non-existence in 1620 holds no water. In the basic
atemporal sense, Brahe's existence just means his existence in 1600,
1590 and in general at all times between 1546 and 1601. Put in more
explicit stage-theoretic terms, what is asserted here is the existence
of Brahe's stages in 1546-1601 and non-existence of his stages at
other times. The same applies to Descartes, Galileo, and others. But
all these cosmologists (i.e., cosmologist stages) coexist with each
other as sharing the same 4D spacetime world, and of course they
coexist with everything else that ever existed, exists or will exist in
the entire history of the universe. All such entities are in the single
domain of quantification of the theory.
I agree that this atemporal concept of sharing the single space-
time manifold has its role to play in the 4D framework. But I think
it becomes rather trivial, hence useless, when applied to stages as
opposed to worms. The basic sense in which complete 4D worms
share the same world, or coexist, is indeed the broad atemporal one:
they all coexist by populating the same 4D spacetime manifold. But
the stage ontology invites a different notion of sharing a world.
Stages are confined to particular times rather than being spread
over intervals of time. The stages' world is a world stage. It is true
but trivial that all stages populate the single 4D manifold and,
hence, all coexist. But that is not the issue with which we are pri-
marily concerned when we talk about, refer to, and quantify over
stages. A more important notion of coexistence pertinent to stages
relates to global stage sharing. In the broad but trivial sense, I (i.e.,
my current stage) share the single world with George W. Bush (his
current stage) and Napoleon (e.g., his 1815 stage). But in our dis-
course we are inclined to draw a distinction between such cases.
And in drawing such a distinction, we are getting at the sense of the
coexistence relation such that I bear this relation to Bush but not to
Napoleon. This non-trivial notion of coexistence implies a correlated
and temporally loaded notion of existence to go along with it. The
latter is brought to the forefront when I state, in all seriousness, that
Bush still exists but Napoleon does so no longer. Obviously, both the
14
This comment was prompted in large part by very useful discussions
with Ted Sider and Mike Rea.
230
 On Stages, Worms, and Relativity

non-trivial notion of coexistence and the accompanying notion of

existence relevant to the stage theory have to do with the sharing
of a single temporal world, or a global world stage, by local
objects.
Yet one might have doubts about the possibility of accommodat-
ing what appear to be two distinct concepts of existence in a single
ontology: the universal concept and the restricted one confined to
particular global stages. To dispose of such doubts, an analogy with
modal realism could be helpful. As noted by Lewis, to quantify cor-
rectly over possibilia and, in general, to do justice to the modal dis-
course, the modal realist needs two different quantifiers: one rang-
ing over the contents of the entire collection of possible worlds and
the other restricted to a particular such world (see Lewis 1986, 3,
5-7, and elsewhere). Any two objects populating the Lewisian
multi-universe coexist in the broad sense, but those belonging to
different worlds do not coexist in the restricted sense. An inhabitant
of our world can state, with ontological seriousness, that alien
objects do not actually exist (meaning their non-existence in our
world). The distinction drawn above between the broad and
restricted concepts of (co)existence in the 'temporal multi-universe'
(i.e., in the complete four-dimensional spacetime manifold, which
can be foliated into families of world stages) parallels the Lewisian
distinction between the two quantifiers. The first ranges over all
stages of all objects while the second is restricted to stage mates—
just as the Lewisian special quantifier is confined to world mates. In
fact, the first distinction has more intuitive appeal, as it allows the
stage theorist to give due respect to familiar differences, such as that
between our coexistence with Bush and our lack of coexistence with
Napoleon, whereas rather esoteric intuitions are required to recog-
nize the realm of possibilia in the first place. Nonetheless, the par-
allel is useful as it brings out, once again, important similarities
between the modal and temporal discourses.
With this distinction in mind, we can provide a stage-theoretic
analysis for much of ordinary discourse about temporally qualified
existence and coexistence. Start with a simple observation that the
1620 stage of Descartes coexists with the 1620 stage of Kepler, but
not with his 1615 stage. The first, but not the second pair of stages
are stage mates, that is, share a single global stage, or temporal
world. Such basic relations among local stages can be used to
analyse many familiar locutions.
• The stage-theoretic analysis of 'In 1620, Descartes coexisted
with Kepler and Galileo but not with Brahe' would go along the

231
Yuri Balashov

following lines: the 1620 stage of Descartes coexists (in the nar-
row sense)—or shares a global stage (a temporal world)—with
some stage of Kepler and some stage of Galileo but with no
stage of Brahe. (Note the tenseless mode of present-tense verbs
in the analysans here and below.)
• As already noted, 'Brahe did not exist in 1620' means that no
stage of him exists then.
• 'Descartes, Kepler, Galileo, and Brahe were contemporaries'
translates into the statement that some stages of Descartes,
Kepler, Galileo, and Brahe are all stage mates, for example, their
1600 stages.
• 'Descartes, Kepler, Galileo, and Newton (1642-1727) were con-
temporaries at some point or other but not all at once' translates
into the statement that some stage of each of them shares a sin-
gle temporal world with some stage of another or with some
stages of some others, but no single global stage contains stages
of all of them.
• 'Neither Descartes nor Kepler was a contemporary of Ptolemy'
means that no stage of Ptolemy shares a temporal world with
any stage of Descartes or Kepler.
Although this framework is transparent enough, some of its features
should be noted, as they will play a role in the subsequent discus-
sion.
1. The relation of coexistence in a single temporal world—the
stage mate relation—is not a set of dyadic relations between mem-
bers of pairs of objects (i.e., their stages), but a single many-place
relation among n-tuples of objects. Descartes, Kepler, Galileo, and
Brahe coexisted in 1600, not because six dyadic relations obtain
between members of pairs of their stages, but because a single four-
place relation obtains among all four.
2. Temporal worlds (global stages) hosting local stages of objects
can be individuated by very minimal amounts of their contents.
Given a single stage of a single object—for example, Descartes in
1620—the global stage representing the whole temporal world con-
taining him is uniquely fixed: the world in 1620. To put it differ-
ently, Descartes' local 1620 stage can be uniquely extended to the
global 1620 world stage.
3. If a collection of stage mates coexist with 01—that is to say, if
the original collection and 01 are related by an n-place stage mate
relation—and also coexist with 02 (in the same sense), then that
collection coexists with both 01 and O2; which is to say that the
original collection supplemented by 01 and O2 are related by an

232
 On Stages, Worms, and Relativity

n+\ -place stage mate relation. Thus Descartes in 1600 and Galileo
in 1600 coexist with Kepler (his 1600 stage); these cosmologists
bear a three-place stage mate relation to each other thereby sharing
the 1600 world. Furthermore, Descartes in 1600 and Galileo in
1600 coexist in the same sense with Brahe (his 1600 stage). Not sur-
prisingly, Descartes in 1600 and Galileo in 1600 coexist with both
Kepler and Brahe by bearing a four-place relation to their corre-
sponding stages and thereby populating the 1600 world. Let us call
this feature of the stage mate relation non-contextuality. The full
meaning of this term will become clear later on.
4. As Descartes grows older he finds himself in temporal worlds
with progressively older stages of Kepler and Galileo. (Regrettably,
some of those later worlds do not contain Brahe.) In this sense, the
stage mate relation is chronologically well behaved.
5. The above analysis is classical. It exploits the intrinsic geome-
try of pre-relativistic (Galilean, or neo-Newtonian) spacetime. The
classical spacetime manifold can be uniquely foliated into a family
of hyperplanes of simultaneity indexed by the absolute time of
Newtonian physics. Every such hyperplane cuts across an entire
temporal world. In fact every hyperplane is a world or, more pre-
cisely, the geometrical locus of a temporal world. Each local stage
finds itself on one of the hyperplanes. And the extension of a local
to a global stage is achieved by drawing a unique hyperplane of
simultaneity through that stage. Figure 1 illustrates these observa-
tions.
Classical spacetime provides a rather friendly environment for
the stage ontology by making the stage mate relation non-contextu-
al and chronologically well behaved. These benefits are easily over-
looked because we normally take pre-theoretical considerations
underlying them for granted. These benefits, however, are lost in
the transition to special relativity. In the remainder of the paper I
shall argue that, on plausible formulations of the stage theory in rel-
ativistic (Minkowski) spacetime, one has to surrender non-contex-
tuality and chronologically good behavior of the stage mate relation.
Before we do it, however, let us turn briefly to the worm view and
indicate that it suggests a different analysis of the situation consid-
ered earlier. On the worm theory, Descartes, for example, is a 4D
object (a 1D world line in Figure 1, in which two spatial dimensions
are suppressed and, in addition, Descartes and others are taken to
be idealized point-like objects) and the same is true of Galileo,
Kepler and others. Although one could sensibly talk about their
stages and various relations among them, this talk is not ontologi-
cally significant in a sense in which the stage theorist's talk of stages
233
Yuri Balashov

is. Ordinary objects, on the worm view, are not stages but worms;
the relations among their stages are derivative from more funda-
mental relations among the 4D wholes. When asked, 'How many
great cosmologists are there?' the stage theorist will take this to be
an incomplete question about cosmologist stages and the appropri-
ate answer will require additional information about the time of
evaluation. There are, for example, three great cosmologists in exis-
tence in 1620 and four in 1600. The basic objects of the stage ontol-
ogy populate global stages; their common world, recall, is a stage.
The basic objects of the worm ontology, on the other hand, popu-
late the entire 4D spacetime reality. When asked, 'How many great
cosmologists are there?' the ontologically serious worm theorist
should regard this as a complete question that is primarily about
perduring wholes, and she will have to list all the great 4D cosmol-
ogists—past, present, and future. The list will include Ptolemy as
well as Hubble, in addition to Galileo, Kepler, Brahe, and Newton.15
Now of course, the worm theorist is free to talk about cosmologist
stages as well, when confronted with the question 'How many great
cosmologists were there in 1620?' And here, she could adopt the
stage theorist's approach to local and global stages and their rela-
tions described above. But far from considering such relations to
pertain to the very essence of existence and coexistence, the worm
theorist should regard them as mere perspectival restrictions of the
more fundamental relations holding among what she takes to be
perduring objects—total 4D entities in the spacetime manifold.
Thus she would say that, from the perspective of 1620, Kepler but
not Tycho coexists with Galileo and Descartes, because the appro-
priate relation obtains between the 1620 stages of Kepler and
Galileo but not between the latter and the 1620 stage of Tycho
(there is no such stage), whereas Galileo and Descartes' common
1600 perspective includes both Kepler and Tycho. These relations
are underwritten by the spacetime relations among the correspond-
ing 4D objects populating a single neo-Newtonian 4D world. As
already noted, in the end all such objects are equally 'world mates.'
It normally behooves the worm theorist, but very rarely the stage
theorist, to adopt the God's eye view of the universe.
To take a spatial analogy, consider a depot in the yard of a huge
museum of natural science, which stations trains bearing the names
of prominent cosmologists and natural philosophers (Figure 2).
The tourists sitting in the middle cars of 'Descartes' and the front
cars of 'Galileo' will see each other's trains and 'Kepler' between
15
Cf. Sider on counting worms in §6 of his 1996 paper.
234
 On Stages, Worms, and Relativity

them (Perspective 1). (Let's pretend that the cars are made of glass
or at least have sufficiently large windows to enable the tourists to
see through.) One could imagine them saying, 'Look, there is
"Kepler" over there. Why didn't they put "Tycho" as well? He was,
after all, just as great!' The tourists in the back cars of 'Descartes'
and the middle cars of 'Galileo,' on the other hand, have a better
sense of historical justice: both 'Kepler' and 'Tycho' are in place
(Perspective 2). But of course, the vision of the tourists is always
perspectivally restricted and, hence, incomplete. T o get the full
sense of historical justice, one would need to take the bird's eye view
of the depot.
These perspectival considerations do not seem especially impor-
tant in the classical case. The news they bring—that the 1620 per-
spective is different from the 1600 one—is reassuring to all (the
stage and worm theorists alike) but not particularly illuminating. As
we shall see shortly, the situation is more complicated in the
relativistic case.

• Perspective 4 "
• '

•
.-

" Perspective 1
-
•

Perspective 2
es"

1
lei

<3^/ io -
U
PQ (U

' P
o
"Ty

P-
•

Figure 2.

4. Stages in Minkowski Spacetime

To make a transition to the relativistic case, one should first of all

take care to replace the notion of temporal stage with a more
235
Yuri Balashov

general and relativistically acceptable notion of temporal-like stage.16

A global temporal-like stage is simply a 3D hyperplane in space-
time. It is similar to a classical momentary global stage in that it cor-
responds to a particular time in a given reference frame; but it
requires two separate indices for its individuation: a frame and a
time in that frame. Two indices, instead of just one, as in the clas-
sical case, are needed because of the absence, in the relativistic
framework, of the frame-invariant concept of time. Time by itself
is not enough to identify such a stage: one has to know in what
frame the time is measured.
Figure 3 features various momentary global 3D temporal-like
stages in Minkowski spacetime. (Since two spatial dimensions are
suppressed in this figure, the stages become ID lines; but one
should not forget that in reality, they are three-dimensional.)

GS

Figure 3. Momentary global stages in Minkowski spacetime. GSj and

GS2 correspond to times in frame (x,t) whereas GSj' and GS2' correspond
to times in (x',tf). The local temporal-like stage S of object O belongs to
both GSj and GSj'.

16
This term was suggested in conversation by Steve Savitt. As will
become clear immediately below, a temporal-like stage is a temporal stage
in a particular reference frame.
236
 On Stages, Worms, and Relativity

In the foregoing, we have drawn useful parallels between Lewis's

possible worlds and classical global stages. Now we should note an
important dissimilarity between the former and relativistic global
stages construed as above. Any two Lewisian possible worlds are
entirely distinct: no object can exist in more than one such world.
On the contrary, any two global temporal-like stages corresponding
to different frames of reference intersect. Consequently, a single
local object can in principle belong to more than one temporal-like
world—that is, to more than one global stage. In fact, unless specif-
ic restrictions are imposed on the construction of global stages from
local ones, any object belongs to an infinite number of worlds.
Just like global temporal-like stages, local ones are also individu-
ated by two indices. This becomes especially important for spatial-
ly extended objects. In Figure 4, global stages GSj and GS 2 contain
two temporal-like stages of the planet Earth. They cut the 4D world
volume of the Earth at different angles in spacetime and intersect
one another along a two-dimensional circle. (Because one spatial
dimension is suppressed in Figure 4, this intersection looks like a
one-dimensional line.)
Such 'crisscrossing' becomes less significant for local temporal-
like stages of idealized point-like objects (remember that this ideal-
ization is adopted throughout most of the paper), such as particle C
world line of C

world
volume
of the Earth
intersection of
j and GS 2

Figure 4. The Earth and its temporal-like stages in Minkowski space-

time.

237
Yuri Balashov

at the center of the Earth (Figure 4). Its temporal-like stages could
in principle be uniquely labeled by just one parameter, for example,
the proper time associated with the rest frame of the particle.

5. Global Stage Sharing in Minkowski Spacetime

Let us now turn to the question of how objects can share a global
stage in Minkowski spacetime. Here one should be open-minded, as
it is difficult to decide in advance what form the stage mate relation
must take in the relativistic framework. It is natural to assume that
the intrinsic geometrical structure of Minkowski spacetime gives us
more flexibility than the rather rigid and unsophisticated geometry
of neo-Newtonian spacetime. Hence, it is best to be maximally
unprejudiced, consider all the options available, and be prepared to
let the chips fall where they may.
For the purpose of discussion, it will be convenient to begin with
a non-starter. Suppose there are just two objects, Oj and O2. Under
what conditions can Oj share its particular stage, say S1^, with some
stage of O2? Clearly, there should be a single global stage—a single
temporal-like world containing S1! and some stage of O2. And there
are an infinite number of such global stages. Is any one of them dis-
tinguished? Not in virtue of the intrinsic structure of Minkowski
spacetime alone. But given this structure and the instantaneous state
of motion of O 1; one global stage hosting S1! is naturally privi-
leged—the one corresponding to the hyperplane of simultaneity in
the rest frame of Oj. In Figure 5, it is GS 1 ^,. As long as we know
the velocity of O1 at the location of S1}, GS 1 ,.^ is uniquely deter-
mined.17
But if the 'global extension' of S1! is defined in this way, it will
not, in general, be a global extension of any stage of O2 and, hence,
will not be a common temporal-like world of Oj and O2. Indeed,
the stage mate relation must be symmetrical, O t and O2 must figure
in its specification on a par. But on the present proposal, this, in a
17
Does the concept of velocity make sense for instantaneous stages? One
might think not, because the specification of velocity requires the specifi-
cation of the object's positions at an infinitesimal but still extended inter-
val of time, whereas an instantaneous stage exists at a single time (in a
given frame). As already mentioned, however, the stage theory adopts a
counterpart view of temporal predication. On that view, an instantaneous
stage S located at position x at time t (in a particular frame) has all sorts of
temporal properties, including such properties as being at x+dx at t+dt.
Thus an instantaneous velocity can be attributed to it.
238
 On Stages, Worms, and Relativity

Figure 5.

great majority of cases, is impossible to satisfy. A global stage such

as GS^ggj, distinguished from the point of view of O 1; will not, in
general, be distinguished from the viewpoint of O2. Oj and O2 can
share a global stage defined as above only if they are at rest with
respect to each other. And such cases are extremely rare. If Oj
moves relative to O2, the global extensions of their stages are 'at an
angle' and thus no global stage can be their common stage (Figure
5). Such objects can never be stage mates and, therefore, can never
populate a single temporal-like world.18
It is clear, on reflection, that linking global stages to the state of
motion of local objects was simply a leftover of the classical view. In
the classical case, two object stages shared a global stage whenever
they were co-present to one another. And co-presence simply meant
existing at the same time, the universal Newtonian time of classical
physics. By abandoning this latter notion—by making time frame-
dependent—special relativity prevents one from associating onto-
18
A point duly appreciated by Sider (2001, §5.8) in his discussion of how
to make the stage view consistent with relativity. Sider's vocabulary and
agenda there are, however, rather different from mine and his proposal to
relativize the truth value of tensed utterances about stages to the rest frame
of the speaker is of no help in handling the ontological issues surrounding
the notion of coexistence in the stage world.
239
Yuri Balashov

logical commitments with coordinate time and simultaneity. But

relativity puts something else in their place: the frame-invariant
notion of spacelike relation, or separation. Two spacetime points are
spacelike related just in case there is a reference frame in which they
are simultaneous. And this existential fact is itself absolute: if it
obtains in any legitimate reference frame, it obtains in every such
frame.
This suggests a proposal to ground the stage mate relation in the
objective relation of spacelike separation between local object
stages. In Figure 5, local stages S1! and S2X of, respectively, Oj and
O2 are spacelike related and that may be enough to ensure that they
are stage mates—that there is a global stage containing both of
them, for example, GS 1 ,.^. In this approach, a common global stage
is not required to be specific to either Oj or O2. It can be just one of
the many legitimate (i.e., spacelike) hyperplanes that could be
drawn through S1} and S2j. That there are many others can be seen
as soon as one recalls that two dimensions of space are suppressed
in Figure 5.
The extension of this schema to more numerous collections of
object stages requires caution. Suppose there are five (or more) local
stages of five (or more) objects, all pairwise spacelike related. In
general, there will be no spacelike hyperplane that can be drawn
through all of them. Indeed, five or more points in a four-dimen-
sional space do not always lie on a single hyperplane.19 But one
should not infer from this that five or more objects could not in gen-
eral be stage mates in Minkowski spacetime. Instead, one should
adopt a more charitable approach to the (temporally-qualified)
notion of coexistence in the stage ontology.
The approach just considered starts with a collection of local
stages and then goes on to assert their coexistence (or the lack of it,
as the case may be) in a single global stage. But there is no more rea-
son to do things in this order in the relativistic case than in the clas-
sical case. In the latter case, we do not begin with, say, the 1620
stage of Descartes, the 1620 stage of Galileo, and the 1615 stage of
Kepler, only to conclude that they do not coexist in any single tem-
poral world. That they don't is, no doubt, good news but it is hard-
ly enlightening. A more sensible strategy is to fix attention on a par-
ticular stage of Descartes, say his 1620 stage, and then pose the
question as to what objects it coexists with. It coexists with Galileo
and Kepler in virtue of bearing a (classical) stage mate relation to

" Just as four or more points in a more familiar three-dimensional space

do not always lie on a single plane.
240
 On Stages, Worms, and Relativity

their corresponding 1620 stages, and it does not coexist with Brahe,
because there is no Brahe stage in 1620. Most objects in our world
come to be and cease to exist. And that is a matter of significance to
us when we say that we (that is, our current stages) coexist with
Bush but not with Napoleon.
Similarly, one should not be obligated to start with an arbitrary
collection of stages in Minkowski spacetime, only to discover that
its members could not (in virtue of the above-mentioned geometri-
cal considerations) be stage mates. Instead, one ought to start with
a particular object of interest O at a certain time t in its life career
(which is none other than proper time intrinsic to the object and
measured in its rest frame) and then pose the question as to what
objects it coexists with. Other objects enter into such a relation of
coexistence with O at t by having stages that bear a single many-
place stage mate relation to each other and to S,, the stage of O at t.
On this approach, O at t turns out to coexist (as it should) with what
we would pre-relativistically count as its 'contemporaries' and not
coexist with any of its 'predecessors.'
The notion of a relativistic 'contemporary' is, of course, different
from its classical counterpart. Classical contemporaries exist at the
same moment of a single time, the absolute Newtonian time. No
such concept is available in the relativistic framework. But there is
a good substitute. Each 'contemporary' of O at t is a certain age, the
age in question being measured by the proper time of that 'contem-
porary.' In the end, this enables all relativistic 'contemporaries,'
including O, to enter into the relation of coexistence with each other
on a par. Thus, the 30-year old Data, the 46-year old Captain James
T. Kirk, and the 65-year old Captain Jean Luc Picard are 'contem-
poraries' of each other: their corresponding stages are stage mates.
On the other hand, Klingon Trevor is the relativistic predecessor of
all of them: none of his stages shares a temporal-like world with any
stage of the first three (Figure 6).
Notably, the same 30-year old Data is also a 'contemporary' of
the 40-year old Captain Kirk and the 55-year old Captain Picard.
This is a consequence of the latitude allowed by relativistic space-
time. It may look surprising but is surely tolerable. I show below
that other features of the relativistic stage mate relation are more
pernicious. Before doing it, however, it may be useful to note that
the relativistic account developed here gives the right result in the
classical limit, which is surely a good sign.
The limiting property of scientific theories constitutes the gist of
the correspondence principle, governing the relationship between an
old and a new theory: the former should be (in some sense) a limit-
241
Yuri Balashov

& 1
Q
I,
^^-"765

30 401

1 o/

s
II Figure 6.

ing case of the latter when the domain of application is restricted to

that covered by the old theory. Thus classical mechanics is a limit-
ing case of relativistic mechanics for velocities much smaller than
the speed of light or, alternatively, in the limit c —> °°. Although the
validity of the correspondence principle has been called in question
by the advocates of incommensurability and related ideas in the
post-positivist philosophy of science, it is still used in the founda-
tions of physics as a helpful heuristic device. (Witness recent dis-
cussions of the classical limit of different interpretations of quan-
tum mechanics.)
Be this as it may, the classical limit of the relativistic stage mate
relation is clearly the old classical relation of existing at the same
moment of absolute time. Applied to Descartes and his ilk—objects
moving with small relative velocities and separated by small dis-
tances—the relativistic analysis simply recovers Figure 1, in which
the outsides of the light cones collapse into absolute hyperplanes of
simultaneity, thereby forcing all global stages (generally inclined at
various angles) to become horizontal. The latitude referred to above
evaporates in this limit: Descartes in 1620 has the uniquely-aged
Galileo and Kepler as his contemporaries.

242
 On Stages, Worms, and Relativity

6. Contextuality

We are, however, interested in precisely those features of the rela-

tivistic stage mate relation that deviate from the classical features.
Some such features are unwelcome. It is now time to investigate
them. Begin with contextuality. Object O at t (i.e., its t-stage, St)
coexists with Oj and with O2 but not with both: there is no single
temporal-like world containing S, and some stages of both Oj and
with O2 (Figure 7). One might wonder, however (and rightly so),
why there should be such a single world, given that O2 is in the
absolute future of Oj. Oj may be Captain Kirk's great-grandfather
and O2 his great-grandson. It would be rather strange for anyone to
be a 'contemporary' of both.

Figure 7.

But the problem is easily reproduced by separating Oj and O2

wide enough in space (Figure 8). There we have a situation in which
Oj does coexist with O2 at a given point of its life career (some of
their stages share a global stage, e.g. GS*), and O at t coexists with
Ox and O 2 but not with both.
In fact, a stronger result obtains. Suppose all three objects are
entities that come to be and cease to exist. Then given their suitable
locations in spacetime, they may coexist pairwise at some point or
other of their life careers: all three may have stages that bear dyadic
stage mate relations to each other. But this does not guarantee that
they coexist all together. The failure of such mutual coexistence
occurs in cases in which no selection of the stages of the three
objects can be unified by a single triadic stage mate relation. Thus
243
Yuri Balashov

o
Figure 8.

Kirk

Figure 9. Kirk coexists with Data: they have stages sharing the global
stage GS D K . The same is true of Kirk and Picard, and of Picard and Data.
But there is no single global stage containing stages of all three.
244
 On Stages, Worms, and Relativity

it may be the case that Data coexists with Captain Kirk, Captain
Kirk coexists with Captain Picard, and the latter coexists with Data.
Taken pairwise, they all share temporal-like worlds with each other.
Taken all together, however, they don't share any single temporal-
like world (Figure 9). The same is true of a quadruple of objects.
Any three of them may coexist but this does not entail the coexis-
tence of all four. (Considerations of space prevent me from illus-
trating such a case.)
This result is easily extended to more numerous collections of
objects. Take a suitably situated group of n objects with finite life
spans, such as Star Trek civilizations, that are all created and
destroyed. Add two more such objects. It may be the case that the
entire initial collection coexists with them taken separately but not
with both taken together—even if those two objects, in addition,
coexist with each other. A more involved case would include, say,
1000 Star Trek civilizations C1( ... C1000 and three more: C*j, C*2,
and C*3. All members of the latter group coexist with each other
and any pair of them coexists with all the members of the first
group, Cj, ... C1Ooo- And yet, all 1003 do not coexist. (Again, I leave
these cases without illustration.)
It may be interesting to find out if coexistence (that is, the tem-
porally qualified coexistence grounded in the many-place stage
mate relation) of all n-1-tuples of objects O 1; O2, • •• On need not
entail the coexistence of the entire collection. But even if this more
general result does not hold, the situation is frustrating enough.
Coexistence in the relativistic stage world turns out to be parti-
tioned in a most peculiar way bearing the mark of contextuality.
Facts about coexistence among members of a collection of objects,
however numerous, become sensitive to what other objects are taken
into account. Such facts do not 'add up' properly.
Contextuality of coexistence in the stage world should not be
confused with the breakdown of transitivity. First, coexistence in a
temporal-like world is a many-place relation, not a dyadic one.
Furthermore, even if consideration is restricted to pairs of objects,
contextuality and the lack of transitivity are distinct features. It is
true that coexistence between members of pairs of objects in the
relativistic stage world is not transitive: coexistence of Oj with O2
and of O2 with O3 does not entail coexistence of Oj with O3.
Contextuality, on the other hand, means that coexistence between
members of all such pairs does not entail coexistence among mem-
bers of the whole triple, Oj, O2 and O3.
Why should the stage theorist be bothered by contextuality, but
not so much by the lack of transitivity? One reason is that transitiv-
245
Yuri Balashov

ity of coexistence fails even in the classical stage world. I coexist

with my father (some of our stages are classical stage mates) and he
coexists with my grandfather. But I don't coexist with my grand-
father. But the classical relation of coexistence is free of contextuality.
If my father, my grandfather, and I coexist pairwise then we coex-
ist all together. We do not find this remarkable because we simply
take it for granted. But for the stage theorist, something important
is at stake here.
Indeed, stages live in stage worlds; their common world is a glob-
al stage. That is what they share—just as objects in Lewis's ontol-
ogy share a particular possible world. In a trivial sense, the latter
also share the entire collection of Lewisian worlds. But that sense is
irrelevant, nothing significant turns on it. All the important features
of that ontology, including the modal properties of objects, are
grounded in the facts about what objects belong to what worlds.
The fact that every object also belongs to the whole collection of
worlds bakes no bread. Similarly, the stage theorist should take the
facts about what object stage belongs to what temporal world—and
what other stages it shares that world with—as the ground of all the
important properties exhibited in the stage 'multi-universe.' These
include the temporary properties of objects, the account of change
they undergo, and the ways they interact with each other. The fact
that all stages of all objects trivially share the single 'multi-universe'
occupying the entire spacetime manifold is simply too coarse-
grained to add anything significant to the picture.
In short, the stage theory must recognize existence in a temporal
(or temporal-like) world and sharing such a world as primary cate-
gories of its ontology. But then it is natural to expect such categories
to obey some reasonable 'calculus.' And they certainly do so in the
classical case, where the stage mate relation is grounded in absolute
simultaneity among local stages. Minkowski spacetime suggests a
bona fide candidate to do a similar job, the relation of belonging to
a single spacelike hyperplane, which adequately recovers its pre-rel-
ativistic counterpart in the classical limit and is, as we saw, partly
successful. The problem with it is that it stumbles upon a simple
rule that is, intuitively, part and parcel of the concept of coexis-
tence: if A coexists with B, B coexists with C, and C coexists with
A, then A must coexist with B and C.
To be sure, these considerations do not go as far as to refute the
stage theory. Rather, they indicate what price this theory has to pay
in the transition to relativity. And that is only part of the price.
Besides being contextual, coexistence in the relativistic stage world
is not well behaved chronologically.
246
 On Stages, Worms, and Relativity

7. Chronology

Let's start with an unproblematic case. The 30-year old Data coex-
ists with the 35-year old Captain Kirk and the 40-year old Captain
Picard: they share a common global stage GSj. As Data grows older
and reaches the age of 35, he happens to coexist with a younger
Captain Kirk, who just turned 32, and a younger, 33-year old
Captain Picard: they share a common global stage GS 2 (Figure 10).

Picard

This, however, is not particularly disturbing. At any moment in

his life, Data belongs to an infinite number of temporal-like worlds.
The fact that one can pick out a chronologically ill-behaved series of
such worlds (e.g., the sequence of GSj and GS2) should not be held
against the stage view, as long as another, chronologically well-
behaved series is available. And it is surely available in the case
under consideration; for example, the series including GSj and
GS 3 . The latter global stage, GS 3 , features the 35-year old Data,
and the correspondingly older Captain Kirk and Captain Picard.
One is not saddled with the unpalatable sequence of GSj and GS 2 ,
because there is no reason to allow one to exploit the latitude inher-
ent in relativistic spacetime frivolously, by sequencing global stages
at will. The availability of chronologically well-behaved series of
247
Yuri Balashov

global stages is all that the Star Trek biographer needs to tell a sen-
sible story about the life careers of the three famous characters and
their relations to each other.
But there are cases where a chronologically well-behaved series of
temporal-like worlds is not available and there is no escape from a
disturbing conclusion that ageing results in being a contemporary of
progressively younger companions. Such a case is represented in
Figure 11.

Picard
Figure 11. SGj and SG2 form a sequence of global stages that is
chronologically well-behaved for Data, Kirk, and Picard. But given the
configuration of these objects in spacetime, any such sequence inevitably
picks out progressively (or rather regressively) younger stages of Trevor.

Here the most one can do is to identify a chronologically well-

behaved series of global stages including the correspondingly age-
ing Data, Kirk, and Picard, for example, SGj and SG2. Adding
Trevor to the picture, however, turns the series into a bad one. As
Data, Kirk, and Picard all grow older, they find themselves in
worlds with the younger and younger Trevor. And that is not the
worst possible scenario yet. With some modifications, one could
make progressively ageing Data, Kirk, and Picard unavoidable con-
temporaries of, first, Chief Trevor, next the 15-year old Cadet
Trevor, then the newly born Klingon baby just named Trevor, and
eventually, Trevor's great-grand-grandfather!
This is surely an unwelcome result. Along with contextuality, it
248
 On Stages, Worms, and Relativity

brings out the difficulties of formulating the coexistence relation in

the relativistic stage world.

8. Discussion

Why is the worm theory not afflicted with the same or similar prob-
lems? Because the worm ontology presupposes a different notion of
coexistence. All 4D worms coexist with each other in the
Minkowski world. This does not mean that one cannot pose tempo-
rally sensitive questions about the coexistence of their various parts.
But the answers to them do not get at the basic sense of coexistence
essential to the worm ontology. They are always restricted to a
particular perspective in spacetime, just as the answers to similar
questions were restricted to a purely temporal perspective in the
classical framework.
The perspective in question is associated with a hyperplane
drawn through the location of a given temporal-like part of a 4D
object. Return to Figure 8 but read it differently now. The figure
represents three 4D objects, O, Oj, and O2—three full-blown spa-
tio-temporal worms, all coexisting together in the single spacetime.
These objects have parts, or stages, confined to spacetime points,
and associated with each such part are infinitely many momentary
perspectives on the single spacetime world. Thus the perspectives
linked to the f-part of O include those containing some part of Oj
and those containing some part of O2. As it turns out, there is no
perspective containing parts of both Oj and O 2 . The f-part of O is
unable to share a perspective with objects Oj and O2 taken togeth-
er. Given an appropriate configuration of the three objects (three
worms, that is), a stronger result obtains (cf. Figure 9). Taken pair-
wise, Data, Kirk, and Picard may share spatio-temporal perspec-
tives with each other. Taken all together, however, they may not.
But that is quite different from being unable to share a world. In a
similar vein, being saddled with a chronologically ill-behaved series
of perspectives is rather benign in a way in which being saddled
with a correspondingly bad series of temporal-like worlds is not.
Every such world constitutes an appropriate habitat for an object-
stage, but not for an object-worm. Indeed, temporal-like worlds are
something in which the objects of the stage ontology live and coex-
ist with each other. The objects of the worm ontology, on the other
hand, live and coexist in the much bigger world, the entire
Minkowski spacetime. Their more fine-grained temporal character-
istics are a matter of perspectivalism, not existence.

249
Yuri Balashov

Perspectivalism of this sort should not be surprising as it is a

general feature of our perception of the world. Tourists sitting in
'Descartes' can look out the window at various angles and thus get
different perspectives on other trains. Some tourists see 'Kepler'
between themselves and 'Galileo' (Perspective 3 in Figure 2),
others don't (Perspective 4). This is anything but surprising and has
nothing to do with the real existence of 'Kepler.'
We can view a house from various vantage points and take its pic-
tures. In general, such pictures look very different; in particular,
they feature different two-dimensional shapes of the whole con-
struction and its different parts. Such representations are inevitably
incomplete, because they are perspectivally restricted, and some do
not 'add up.' One of them may feature the kitchen and the living
room, another the kitchen and the bedroom, and the third the
kitchen and the bedroom. There may be no representation featuring
all three rooms together, but that is just what is to be expected. In
fact, nothing could be more natural than such incompleteness and
'contextuality' inherent in spatial perspectivalism. And of course,
these features have nothing to do with existence. Every one knows
that the real house out there is three-dimensional, and its 3D invari-
ant configuration, complete in every detail, stands behind all partial
2D representations.
Similarly, the objective 4D configuration of worms in the single
spacetime stands behind all stage-restricted perspectives. In this
sense, it may not be too far off the mark to say that worms are, after
all, primary and stages secondary.

References

Balashov, Yuri 1999. 'Relativistic Objects,' Nous 33 (1999): 644-62.

2000a. 'Enduring and Perduring Objects in Minkowski Space-Time,'
Philosophical Studies 99, 129-66.
2000b. 'Relativity and Persistence,' Philosophy of Science 67,
S549-S562.
2000c. 'Persistence and Space-Time: Philosophical Lessons of the
Pole and Barn,' The Monist 83, 321-40.
Brogaard, Berit 2000. 'Presentist Four-Dimensionalism,' The Monist 83,
341-56.
Callender, Craig 2000. 'Shedding Light on Time,' Philosophy of Science
67, S587-S599.
Carter, W. R. and Hestevold, H. S. 1994. 'On Passage and Persistence,'
American Philosophical Quarterly 31, 269-83.
Craig, William Lane 2000. The Tenseless Theory of Time: A Critical
Examination. Dordrecht: Kluwer.
250
 On Stages, Worms, and Relativity

Craig, William Lane 2001. Time and the Metaphysics of Relativity.

Dordrecht: Kluwer.
Haslanger, Sally 1989. 'Endurance and Temporary Intrinsics,' Analysis
49, 119-25.
Hawley, Katherine 2002. How Things Persist. Oxford: Clarendon Press.
Heller, Mark 1990. The Ontology of Physical Objects: Four-dimensional
Hunks of Matter. Cambridge: Cambridge University Press.
Hinchliff, Mark 1996. 'The Puzzle of Change,' in J. E. Tomberlin, (ed.),
Philosophical Perspectives 10. Oxford: Basil Blackwell, pp. 119-36.
Johnston, Mark 1987. 'Is There A Problem About Persistence?'
Proceedings of the Aristotelian Society, Supplement 61, 107-35.
Lewis, David 1986. On the Plurality of Worlds. Oxford: Blackwell.
1988. 'Rearrangement of Particles: Reply to Lowe,' Analysis 48,
65-72.
Lombard, Lawrence 1999. 'On the Alleged Incompatibility of Presentism
and Temporal Parts,' Philosophia 27, 253-60.
2000. 'The Lowe Road to the Problem of Temporary Intrinsics,' pre-
sented at the Pacific Division Meetings of the APA (April 2000).
Lowe, E. J. 1988. 'The Problem of Intrinsic Change: Rejoinder to Lewis,'
Analysis 48, 72-77.
Markosian, Ned (forthcoming), 'A Defense of Presentism,' in Dean
Zimmerman (ed.), Oxford Studies in Metaphysics, vol. 1.
Mellor, D. H. 1981. Real Time. Cambridge: Cambridge University Press.
Merricks, Trenton 1994. 'Endurance and Indiscernibility,' Journal of
Philosophy 91, 165-84.
1995. 'On the Incompatibility of Enduring and Perduring Entities,'
Mind 104, 523-31.
1999. 'Persistence, Parts, and Presentism,' Nous 33, 421-38.
Prior, Arthur N. 1970. 'The Notion of the Present,' Studium Generale 23,
245-48.
Quine, W. V. O. 1950. 'Identity, Ostension, and Hypostasis,' Journal of
Philosophy 47', 621-33.
1987. 'Space-Time,' in Quiddities. Cambridge, MA: Harvard
University Press, pp. 196—199.
Rea, Michael 1998. 'Temporal Parts Unmotivated,' The Philosophical
Review 107, 225-60.
Saunders, Simon (forthcoming), 'Why Relativity is Contrary to
Presentism,' in Craig Callender (ed.), Time, Reality, and Experience.
Cambridge: Cambridge University Press.
Savitt, Steven 2000. 'There's No Time Like the Present (in Minkowski
Spacetime),' Philosophy of Science 67, S563-S574.
Sider, Theodore 1996. 'All the World's a Stage,' The Australasian Journal
of Philosophy 74, 433-53.
2000. 'The Stage View and Temporary Intrinsics,' Analysis 60,
84-88.
2001. Four-Dimensionalism. An Ontology of Persistence and Time.
Oxford: Clarendon Press.
251
Yuri Balashov

Taylor, Richard 1955. 'Spatial and Temporal Analogies and the Concept
of Identity,' The Journal of Philosophy 52, 599-612.
Van Inwagen, Peter 1981. 'The Doctrine of Arbitrary Undetached Parts,'
Pacific Philosophical Quarterly 62, 123-37.
1990. 'Four-dimensional Objects,' Nous 24, 245-55.
Zimmerman, Dean (1998. 'Temporary Intrinsics and Presentism,' in
Peter van Inwagen and Dean Zimmerman, (eds), Metaphysics: The Big
Questions. Oxford: Blackwell, pp. 206-19.

252
On Becoming, Cosmic Time and
Rotating Universes1
MAURO DORATO

Abstract

In the literature on the compatibility between the time of our expe-

rience and the time of physics, the special theory of relativity has
enjoyed central stage. By bringing into the discussion the general
theory of relativity, I suggest a new analysis of the misunderstood
notion of becoming, developed from hints in Godel's published and
unpublished arguments for the ideality of time. I claim that recent
endorsements of such arguments, based on Godel's own 'rotating'
solution to Einstein's field equation, fail: once understood in the
right way, becoming can be shown to be both mind-independent
and compatible with spacetime physics. Being a needed tertium quid
between views of time traditionally regarded as in conflict, such a
new approach to becoming should also help to dissolve a crucial
aspect of the century-old debate between the so-called A and B
theories of time.

1. Introduction: the shift from STR to GTR and the

centrality of becoming

In the literature on the relationship between the time of our expe-

rience and the time of physics, the special theory of relativity (STR)
has curiously but undoubtedly played a major role. On the assump-
tions that
(i) becoming (the 'flow of time') is the essential feature of expe-
rienced time;
1
I thank J. Butterfield, C. Callender, R. Clifton, J. Faye, M. Piazza and
the audience at LSE and Vancouver for helpful comments and suggestions
on previous drafts of this paper. S. Savitt deserves a special mention, for
our frequent exchange via e-mail helped me to formulate my views in a
clearer way. Despite some criticism that here I raise to his previous,
thought-provoking work, he has now independently come to defend views
about temporal becoming that are very close to mine, as is evident from the
paper in this collection.
253
Mauro Dorato

(ii) objective (i.e. mind-independent) becoming presupposes an

ontological difference between present and future events or
state of affairs;
(iii) the geometrical structure presupposed by STR is a necessary
constraint that physical time in general must meet,
a solution to the problem of the definability of becoming in
Minkowski spacetime has also been regarded as the main way to
solve the question of the compatibility between the time of physics
and the time of our experience.2
However, while (i) and (ii) above can be regarded as plausible, (iii)
should strike us as suspicious, especially when it is used to claim
that if Minkowski spacetime cannot make room for any sort of
'ontological difference' mentioned in (ii), then becoming must be
considered to be mind-dependent. An endorsement of the truth of
the antecedent of this conditional is usually assumed to have conse-
quences also for the philosophy of time one should adopt. For
instance, to the extent that a commitment to a mind-independent
becoming is regarded as the essential tenet of the so-called A (or
'dynamic') theories of time, those of their B ('static') rivals that
treat past, present and future events as being ontologically on a par
would be vindicated by the geometrical requirements of Minkowski
spacetime.3 On the contrary, if some sort of primitive relation of
becoming—appropriately relativized to points or worldlines—could
be defined in terms of the structure of Minkowski spacetime, the
compatibility between becoming and STR would be demonstrated,
and no choice between the A and the B theories of time would be
possible only on the basis of physics.
To an unbiased reader, however, such an exclusive worry with
STR should appear as puzzling, and in need of a justification. True
enough, Minkowski spacetime is the standard, flat spatiotemporal
arena for contemporary quantum field theories. However, since in
the presence of gravitating matter STR does not yield an accurate
description of physical reality, it cannot be viewed—as (iii) obviously
2
See Rietdijk [1966], Putnam [1967], Stein [1968, 1991], Weingard
[1972], Godfrey-Smith [1979], Maxwell [1985], Dieks [1988], Clifton and
Hogarth [1995], Dorato [1996, 2000], Rakic [1997], Tooley [1997], and
Savitt [2000] among others.
3
For a recent survey on the debate between the A and the B theories of
time—whose formulation dates back to McTaggart [1908]—see Le
Poidevin [1998]. For reasons that will become clear in the following, rather
than referring to the debate by using the misleading terms 'tensed' and
'tenseless' theories of time, I prefer the more neutral 'A' and 'B' theories of
time.
254
 On Becoming Cosmic Time and Rotating Universes

presupposes—as a fundamental physical theory. Considering that

within the general theory of relativity (GTR), STR has only a 'tan-
gential' validity,4 why should we assume that the properties of time
that are characteristic of the latter theory also apply to the former?
More generally, why should we assume that time has the same proper-
ties across different physical theories?
Notably, Weyl [1918, p. 220], Eddington [1920, p. 163], Jeans
[1936] and Godel [1949a] were all aware of the fact that the special
relativistic limitations vis a vis the absence of a distinguished, global
time order can be regarded as a 'local phenomenon'. In fact, the pres-
ence and the actual distribution of matter in the large-scale struc-
ture of the universe may «largely destroy the equivalence of differ-
ent observers, and distinguish some of them from the rest, namely
those which follow in their motion the mean motion of matter»
(Godel [1949a, p. 559/1990, p. 204]). Unfortunately, in much of the
recent literature such an important point seems to have been lost.5
One of the main aims of this paper is to redress the balance, by relo-
cating the discussion about becoming and physical time in the more
appropriate context of G T R and of cosmological models in gener-
al.6
Besides having being too absorbed by the infinitesimal, 'tangen-
tial' features of Riemannian spacetimes, I think it is fair to add that
the philosophical literature has never yielded a clear and convincing
analysis of the rather obscure notion of becoming, something which
has contributed to generate a widespread—but, in my opinion,
totally ungrounded—belief in the incompatibility between the time
of physics and the time of our experience. Such an obscurity has
also affected the formulations of the two major theories that have
divided the analytic 20 th century philosophy of time. As it is should
be clear from the above presentation, I take it that the real con-
tention between the 'A' and the 'B' theories of time does not concern
4
The pun of 'tangential' refers to the validity of STR in planes that are
tangent to each point of a Riemannian manifold of GTR. The pun is in
Savitt [2000].
5
Saunders [1996] offers a brief discussion of cosmic time in the context
of Godel's argument, and defends a relational view of tenses with which
this paper is in complete agreement, though he would probably disagree
with the view of becoming presented here. Yourgrau's [1991], Savitt's
[1994] and Earman's [1995] contributions will be discussed below.
6
Of course, GTR might end up being a phenomenological, derived
theory as well. However, until a reasonably agreed upon quantum theory
of gravity is available, we can assume that GTR is a fundamental physical
theory.
255
Mauro Dorato

the truth conditions of tensed sentences (as Faye [1989] and Mellor
[1998] have it), or the relational versus the monadic nature of tens-
es (as Horwich [1987] among others has it). In spite of the obvious
importance of these questions, in the following I will take for
granted that the crucial, still open rift between the two camps con-
cerns the nature of change and the mind-independence of becoming.1
Especially within the recent attempts at grounding a quantum
theory of gravity, time seems to have lost the independence it had
acquired with respect to change in the complex historical path that
led from Plato to Newton (see Smith [2000, p. 928-9]), and today it
seems appropriate to regard the notion of time as being inextricably
linked to that of change.
Given the importance of a correct understanding of becoming
for this project, I plan to begin by proposing a new analysis of such
a notion, to be regarded, on the wake of Godel [1949a], simply as
the successive occurrence (coming into being) of tenselessly conceived
facts or events (§2). Armed with such a much needed tertium quid
between the traditional ontological requirements of the A and the
B theories of time, I will then show that both Godel's argument
against the reality of time based on his famous 'rotating universes',
and its recent endorsement by Savitt [1994], fail (§3 and §4).
Despite the fact that physics in principle cannot yield a sufficient
condition for the tenseless coming into being of events at instants of
cosmic time which becoming consists in, I claim that the cosmolog-
ical model currently adopted by physicists is completely consistent
with it and with the requirements of experiential time, once the lat-
ter has been correctly explicated. Finally, by showing that my
explication of becoming is faithful to our pre-theoretical intuitions
about it and does not run into notorious paradoxes entailed by 'the
moving now', I conclude with a simple argument in favour of its
objectivity (§5).

2. The nature of becoming and Godel's argument for the

ideality of time

Godel's argument against the reality of time, which appeared in

Schilpp's volume in honour of Einstein (Godel [1949a]), is based on
7
Here I follow Tooley [1997], who has convincingly argued that grant-
ing (as I do) (1) that the truth-conditions of tensed sentences are given by
tenseless sentences and (2) that tenses are relations, does not yet solve the
problem of becoming and of the ontological status of future events, which
is what I am after here.
256
 On Becoming Cosmic Time and Rotating Universes

the discovery of a new solution to Einstein's field equation, notori-

ously encompassing the existence of closed timelike curves (Godel
[1949b]). The argument is important not only for the conclusion
it—unsuccessfully, as we will see—tries to support, but much more
for the brilliant analysis of controversial philosophical notions that
it provides; from this point of view, it has certainly not received the
attention it deserves.8 Besides Yourgrau's pioneering work on
Godel's philosophy of time [1991, 1999], which had the great merit
of taking into account also Godel's unpublished material, there are
as of now two conflicting reconstructions of Godel's argument for
the ideality of time, Savitt's [1994] which endorses it, and Earman's
[1995, pp. 194—200], which rejects it, and somehow considers it
unworthy of much attention.
Part of the neglect of this argument in comparison to the ques-
tion of time travel, also raised by Godel's model, can be explained
by the fact that Godel's argument is incomplete and 'gappy' to say
the least, as it appears to be centred around the cryptic claim that
since there is no objective lapse of time in his rotating universe, there is
no objective lapse of time in our world either, the main difference
between the two models depending only on the way matter is con-
tingently distributed and moves. More specifically, in our universe,
unlike Godel's, matter is not everywhere rotating (as Godel put it,
«the compass of inertia does not rotate around galactic matter»),
though the physical laws given by Einstein's equations are the same, as
Godel's model satisfies them. Interestingly, Godel discovered that
the lack of rotation is sufficient to define a global temporal order
(see Malament [1995, p. 263]), since the congruence of worldlines of
matter corresponding to the major mass points of the universe can
be compared to the strands or the fibres of a rope representing
spacetime. Absence of twisting, which corresponds to null rotation,
is sufficient to slice through the rope with a plane which is orthog-
onal to every fibre of the rope and the collection of all such planes is
called 'cosmic time'.
To philosophers of space and time, it is indeed reassuring to find
out that Godel's interest in general relativity was philosophical in
origin, as his mathematical work on time «was spurred by his inter-
est in Kant's philosophy of space and time rather than by his fre-
quent talks with Einstein», which in any case began only in 1942
(Wang [1995], p. 216). In fact, in his 'Lecture on rotating universes'
[1990, p. 274], Godel himself tells us that he was motivated to find
8
Though Stein [1970] had already stressed its philosophical signifi-
cance.
257
Mauro Dorato

his new 'rotating solutions' to Einstein's field equation to rebut an

argument due to Jeans [1936], in which it is maintained that the
general theory of relativity has re-established the possibility of an
'objective lapse of time'.9
In order to thoroughly understand the argument I am about to
present below, two terminological points are appropriate. First, it is
important to keep in mind that the notion of change that Godel
introduces in the argument is at variance with much of the analytic
tradition in the philosophy of time, since it requires an objective
coming into existence of facts or events (this coming into existence he
calls: 'the lapse of time'). While within such a tradition change pre-
supposes just the possession of two incompatible properties exem-
plified by the same perduring entity at two different times, Godel's
notion of change is tantamount to an objective coming into being,
and is to be regarded as an essential feature of the time of our expe-
rience. In a passage in the manuscript B2, where Godel summarizes
the result of his investigation into the structure of time in STR
[1995, p. 236], he writes «what remains of time in (special)
relativity theory as an objective reality inherent in the things neither
has the structure of a linear ordering nor the character of flowing or
allowing of change. Something of this kind, however, can hardly be
called time (my italics))). In other words, according to Godel, time is
real only if both a linear ordering and an objective lapse exist indepen-
dently of observers.
The second remark is that in the published piece [1949a], he
defends also the converse claim that change presupposes an objec-
tive lapse of time. These two claims together imply that time is real
if and only if a change in the existing is real. This equivalence elim-
inates the charge of circularity in the first three premises of the
argument below, and justifies in particular its first premise (0),
which in the published paper has no textual support, but is obvi-
ously assumed for the sake of the conclusion about the ideality of
time.
Whenever possible, each premise of the argument—whose recon-
struction owes much to both Earman's and Savitt's—is supported
by textual quotations from Godel's published work [1949a]. Partial
conclusions deduced from previous premises are in bold types:

9
The English word used by Godel, 'lapse' comes from the Latin labi,
which means toflow.So lapse of time is equivalent toflowof time, in the
way to be clarified below.
258
 On Becoming Cosmic Time and Rotating Universes

GODEL'S ARGUMENT AGAINST THE REALITY OF TIME

Parti
(0) Time is real only if change is real.
(1) Change is real only if there exists an objective lapse of time.
«change becomes possible only through the lapse of time»
(1949a, p. 558/1990, p. 202)

(2) Time is real only if there exists an objective lapse of

time [from (0) and (1)]
(3) «The existence of an objective lapse of time means or at least
is equivalent to the fact, that reality consists of an infinity of
layers of 'now' which come into existence successively))
(1949a, p. 558/1990, p. 202).
(4) Reality consist of an infinity of layers of 'now' which come
into existence successively only if spacetime admits of a glob-
al time function {cosmic time).
(5) Time is real only if spacetime admits of a global time
function [from (2), (3) (4)]
(6) Godel's rotating-model M, qua solution to Einstein's field
equations, is a physically possible model, and despite the pres-
ence of closed timelike curves (circular time) and looming
grandfather paradoxes, cannot be ruled out a priori.
(7) Since for every x in M, x chronologically precedes itself, M
does not possess a global time function.
(8) In the physically possible world M, time is ideal
[from (5) (6) (7)]

Part II

(9) The main, contingent, non-lawlike difference between M

and our universe is given by the (probable) absence of a net
rotation of matter, which implies the existence of cosmic
time in our world
(10) ?

(C) Time in ideal also in our universe

Two obvious questions must be answered in order to see whether

Kant's theory of an ideal (transcendental) time is really vindicated
by Godel's rotating universes, as the Austrian logician had it: (i) is
the first part of the argument valid? (ii) if it is, how do we bring its
259
Mauro Dorato

conclusion to bear on the status of time in our universe, which does

not seem to show any rotation of the kind required by Godel's
model (the second part of the argument)? The second question is
clearly linked to the problem of filling the premise (10).

3 The first part of Godel's argument

The unanimous opinion of commentators is in favour of the con-

clusion of part I of the argument, which proves that in Godel's uni-
verse time is ideal, or mind-dependent. However, some of the
premises in my reconstruction, which differs from Savitt's and
Earman's, might be regarded as controversial.
For instance, and firstly, it could be objected that (1)—and there-
fore (2)—are not plausible, as they imply the dubious theory of an
absolute change in what exists, rather than an ordinary, qualitative
change of what already (tenselessly) exists, as in 'the party became
boring' or 'the traffic light became red'. Absolute change in this sense
is what Godel called 'a change in the existing', already distinguished
from qualitative change by C. D. Broad long ago: «To 'become pre-
sent' is, in fact, just to 'become', in an absolute sense; i.e., to 'come
to pass' in the Biblical phraseology, or, most simply, to 'happen'.
Sentences like 'This water became hot' or 'This noise became loud-
er' record facts of qualitative change. Sentences like 'This event
became present' record facts of 'absolute becoming')) [Broad 1938,
p. 280]. To counter this first objection to Godel, it is then impor-
tant to keep in mind that 'change' as used in premise (0) refers to
absolute change (absolute becoming) in Broad's sense, to be carefully
distinguished from a qualitative change of events losing the (pseudo-
attribute of) 'being future' and becoming present.
Secondly, it might be objected that (2) implies the dubious
'moving now' conception of time (see Earman [1995] and Savitt
[1994, p. 468]), since it is always possible to ask 'how fast does the
absolute change in what exists occur?' However, as anticipated ear-
lier, I argue that the claim that instantaneously conceived events (or
facts) 'come into existence' at a certain time (the 'objective lapse of
time' in the above argument) is simply equivalent to the claim that
they mind-independently occur at that time. Consequently, Godel's
locution 'events come into existence successively' should really be
read simply as 'events (mind-independently) take place one after the
other at their time of occurrence'.
In a word, as I interpret it, the objective lapse of time or the
'change the existing' referred to by Godel amounts to the rather non-
260
 On Becoming Cosmic Time and Rotating Universes

metaphysical, almost self-evident claim that if 'event E occurs (or,

equivalently, tenselessly exists) at time t', at a later or earlier time t",
other events occur (exist)10. This means that, at time t', the set of
existing events includes events other than those existing at t". With
this stipulation, our language regimented in a logical way would
have a domain of quantification for each time, containing only those
entities that then exist.
To come now to the charge that an objective coming into being,
or the 'flow (lapse) of time' as it is interpreted here, implies the fal-
lacious 'moving now', note that one could simply point out that the
absolute change in what exists is no ordinary change, and as such, it
does not conceptually depend on other notions as the latter does." The
reason why it is simply meaningless to ask 'how fast does such a
change in the existing occur?' is given by the fact that the notions
that are synonymous of becoming or 'coming into existence', namely
'occurring' or 'happening', are not further analysable; in any case,
they don't presupposes a perduring entity and a pair of incompati-
ble properties possessed by the same entity at different times as the
qualitative notion of change does. It is only in the ordinary sense of
change—the qualitative change of, say, a piece of iron becoming
rusted—that one can talk about the rate of change, since any change
in time can be slow or fast (a slow aging or rusting, a fast aging or
rusting). Of a change of time, one cannot even say that it occurs—
though it can be regarded it as a feature of the universe quite inde-
pendent of our minds—since, strictly speaking, it is only events that
can occur at times, and their succeeding one another at different
times is not an event, if the latter is defined as an instantaneous enti-
ty as is customary in relativity.
A third objection a tenseless theorist of time might have against
(1) above is that it ignores the tenseless aspect of time. According to
the tenseless theorist, events are mind-independently before one
another, even though they are given in block, because they don't
become, or don't come into being or cease to exist {the block uni-
verse). In a word, for certain B theorists like Mellor and Faye, tense-
less temporal relations—and therefore, in a sense, time—are real
even if the lapse of time usually advocated by the A camp (i.e., the
coming into existence of events) is mind-dependent, so that (2) is
false.
10
11
'Exist' here is meant in a tenseless sense, given by 'existing at a time'.
This is the line also taken by Savitt in his contribution to this volume.
As a matter of fact, we arrived independently at the importance of care-
fully distinguishing absolute change from ordinary qualitative change in
Broad's sense.
261
Mauro Dorato

As a reply to this third criticism, note that even if we changed (2)

above by requiring that
(2)' Time is real only if the distinction between before and after
is mind-independent (objective),
a tenseless theorist would still have a harsh destiny in M. Given the
existence of closed timelike curves for any point of GodePs spacetime,
an observer whose spatiotemporal carrier coincided with a segment of
such curves would have no justification for claiming that beforeness or
afterness is mind-independent. Events of type E that she would expe-
rience as being before events F, on a closed timelike curve would also
be such that F is before E, so that, in such a Godelian world, temporal
betweenness would seem the only objective relation «inhering in
events*. Consequently, as Kant had it, in GodePs universe it would be
plausible to assume that time as we experience it emerges from the
relation of our faculty of perception with the «things in themselves*,
which established the conclusion of the first part of the argument.
In order to give further arguments in favour of premise (2), it is
of paramount importance to keep in mind that when Godel refers
to 'time', he always means 'the time of our experience', or «what
everybody understood by time before relativity theory existed*
(1990, manuscript C l , p. 247). In particular, this implies that, in any
case, premise (1)—and (2)—do not purport to say something about
physical time or the metaphysics of time in general, but only about men-
tal, experienced time. Considering that the overarching purpose of
Godel's paper is to re-evaluate Kant's theory of time and show that
it is not only compatible with relativity but even vindicated by it—
as is also clear from the opening paragraphs of the two manuscripts
preceding [1949a]—premise (1) needs no justification from the
moving-now conception of time, as Earman speculates [1995,
p. 199]. Premise (1) is assumed only to prove that if spacetime does
not make room for a necessary condition for objective (mind-inde-
pendent) coming into being, namely cosmic time, Kant's thesis
about the ideality of time would be correct, against the prevailing
opinion of 20th century philosophers of space and time.12
12
'Prevailing', however, does not mean all: witness the contemporary
theoretical physicist Rovelli, and the way he concludes his overview of the
problem of time in quantum gravity: «If time is the order of the changes in
the states of the systems, and if the state of a system is a relational notion,
one that has meaning only if referred to an observer, can there be time out-
side the observer/observed relation? Is perhaps time precisely what
emerges from this observer/observed relation? Is time precisely such a
relation?*. Rovelli [1997, p. 217].
262
 On Becoming Cosmic Time and Rotating Universes

A fourth controversial point of the first part of GodePs argument

might concern the condition of globality: one could object that in
the spacetime of general relativity, such a condition may not be nec-
essary to the existence of a lapse of time, and therefore question
premise (4) above. One could conceive a local, mind-independent
coming into being along single worldlines also in a Godelian uni-
verse, not matched by analogous phenomena at a cosmic scale.
Likewise, the absence of an invariant, global time order in STR
could be compensated by a worldline-dependent becoming, as is
proposed by Clifton and Hogarth [1995].
In the same fashion, for example, Boltzmann thought that the
universe could be in a global state of thermal equilibrium, while
some regions, large as a cluster of galaxies, could be characterized
by gigantic, rare fluctuations, due to which, for some billions of
years, observers would reckon an increase of entropy, and therefore
some sort of objectively irreversible phenomena (Boltzmann [1896-
98/1964]). Would we deny that entropy grows in those regions sim-
ply because at a larger scale, both spatially and temporally, the uni-
verse is in equilibrium? I doubt it. But then, why can't we say that
some sort of local becoming takes place in a mind-independent
way?
Godel would probably object that by admitting a local coming
into being, where 'local' here has the same sense it had in
Boltzman's 'pockets of increasing entropy lasting for eons', we
would make a change in the existing—the lapse of time—relative to
particular worldlines, i.e., to some possible observers living in a galaxy.
And then, he would add: «The concept of existence (...) cannot be
relativized without destroying its meaning completely* [1949a
559/1990, 223, fn. 5].
However, note that if we cannot relativize the concept of exis-
tence, an examination of the impact of the special and the general
theory of relativity on our ordinary notion of time would be mean-
ingless since, independently of relativistic considerations, such a rela-
tivization is implicit in the very idea of a lapse of time even accord-
ing to Godel. In our experience, as he wrote, we often assert of the
same event that «it exists and it does not exist, at two different instants
of time». Furthermore, without such a relativization, we would be
subject to some form of McTaggart's paradox about events being
present (existent) and non-present (non existent) at the same time.
These remarks are of paramount importance, since not only do they
entail that a relativization of tenses is necessary, but also that it does
not lead us to a view of becoming that is too deflated to be worth
having (see Callender [1997, p.118]). Elsewhere (Dorato [1995]), I
263
Mauro Dorato

have argued that there cannot be a future event in an absolute sense,

since an event can count as future only relatively to some present
event or other, and human existence appears always temporally
located and perspectival, that is, experienced at each instant of time
from the perspective offered by that instant. Here, let it suffice to say
that, beyond the possibility it offers of re-establishing a compatibil-
ity with physical time, the main reason why one wants to defend such
a perspectival, relational understanding of existence in time is that
without it we could not make room and explain our capacity to liter-
ally bring about a future event by acting in the present: 'making things
happen' presupposes that events that are yet to occur and are
brought about by our efforts do not (tenselessly) exist relatively to
the moment of action. If they did, our action and our experience of
passage would be both illusory, and utterly unexplainable.
Granting the possibility of relativizing the concept of existence
in this sense, a much more plausible defence of the condition of
globality is that, by rejecting it at least in the context of Godels's
spacetime, we would make the lapse of time non-intersubjectively
valid: «in whatever way one may assume time to be lapsing there will
always exist possible observers to whose experienced lapse of time
no objective lapse corresponds (in particular possible observers
whose whole existence objectively would be simultaneous))) (Godel
1949a 561/ 1990, 205-6). Godel here refers to hypothetical
observers Oj, whose worldlines lie beyond a certain critical point P
of his spacetime model, characterized by the fact that the light
cones at P are tangent to the hyperplane of simultaneity determined
by those observers (call them O) that are located in the convention-
ally chosen axis of rotation of Godel's universe. Since, beyond P,
Oj's closed worldlines belong to a hypersurface of simultaneity deter-
mined by O, Oj's whole existence along the circular time-like curve
would be simultaneous with a particular instant in O's existence.13 In
view of this peculiarity of Godel's spacetime, I take that in the con-
text of the argument under discussion it is plausible to grant
Godel's implicit condition of globality, in such a way that an objec-
tive lapse of time must be a lapse for all possible observers (world-
lines) of the spacetime. In a word, making a reasonable 'equation'
between a possible observer and a worldline, within Godel's cosmo-
logical model the objectivity of becoming must imply its intersubjective
validity.
We can therefore conclude that if we lived in Godel's universe,
13
For a vivid representation of this situation, I refer the reader to the
picture in Malament [1985]. See also Savitt [1994, note 10].
264
 On Becoming Cosmic Time and Rotating Universes

we should be Kantian about time, since both the difference between

earlier and later and that between present and future would be
mind-dependent.

4 The second part of Godel's argument: why the epistemic

defence fails

How does the valid conclusion of the first argument impinge on the
way we should understand time in our universe, where the distrib-
ution of matter is different? There are two possible interpretations
of Godel's argument, an epistemic one and a metaphysical-modal
one, pointing to the necessary grounding of cosmic time in the laws
of nature. Here I will limit myself to the former interpretation,
which is essentially due to Yourgrau [1991] and Savitt [1994]. Not
only is this choice motivated by the fact that it has generated more
discussion than the latter, but also by the remark that the meta-
physical interpretation has unanimously been regarded as being
extremely difficult to justify.
Suppose, with Savitt [1994], that in a physically possible Godelian
model, there are inhabitants like ourselves measuring a local time t^
in the local «compass of inertia», in such a way that whenever x tem-
porally precedes y for any two events in the galaxy where the
Godelians live, tL(x) < tL(y). Then it could be argued that the direct
experience of time of the Godelians is exactly like ours. On the basis
of this remark, Savitt has thus reconstructed Godel's reasoning:
(10) it is possible to have direct experience of time just like ours
in a universe in which (as in M) there is no objective lapse of
time; [recall (8)]
(11) such an experience provides the only reason to suppose that
there is an objective lapse of time in our universe;
(12) «our direct experience of time provides no reason to
suppose that there is an objective lapse of time in our
universe» [from 10 and 11]
(13) «Since there is no objective lapse of time in M, there is
no reason to suppose that there is an objective lapse of
time in our universe* (Savitt [1994, p. 468]).
[from 8 and 12]
This reconstruction has the undeniable merit of being faithful to
the text, as it is probably spells out what Godel had in mind when

265
Mauro Dorato

he wrote: «if the experience of time can exist without an objective

lapse of time, no reason can be given why an objective lapse of time
should be assumed at all.» (Godel [1949a, p. 561/1990, pp.
205-206]).
Such an epistemic interpretation of Godel's argument had
already been anticipated, somewhat more concisely but less per-
spicuously, by Palle Yourgrau: «Since the actual world is lawlike
compossible with the Godel universe, it follows that our direct
experience of time is compatible with its ideality (assuming with
Godel, its ideality in the Godel universe). But if even direct experi-
ence is inadequate to establish the existence of intuitive time—that
is, not merely (relativistic) causal or cosmic time, but genuine, suc-
cessive time that lapses or passes—then nothing further will suffice*
(Yourgrau [1991, p. 53]). In a word, Yourgrau-Savitt's epistemic
argument weakens Godel's attempted conclusion, as it amounts to
shifting the burden of proof to the defenders of the reality of the
time of our experience.
The latest attempt at an evaluation of the gist of this argument is
Earman's, who, in the appendix to the chapter 6 of his [1995], exam-
ines Yourgrau's version as is reported above and rejects it—he does
not discuss Savitt [1994], as the paper was probably in press. Earman
tells us that «apart from our experience, we have all sorts of evidence
that lend strong support to the inference that we do not inhabit a Godel
type universe, but rather a universe that fulfills all of the geometrical
conditions necessary for an objective lapse of time.)) [1995, p. 199].
Unfortunately, it seems to me that Earman has misconstrued
Godel's argument and Yourgrau's main point. The crux of Godel's
argument is not that our scientifically tutored experience, together
with inferences to theoretical structures, does not suffice to establish
that we live in a universe endowed with cosmic time, as Earman seems
to have it. Rather, Godel's point, as correctly reconstructed by
Yourgrau and Savitt, is that after the discovery of the rotating solu-
tions to Einstein's field equation, our experience alone (without the
help of independent arguments) is not sufficient for objective becom-
ing, i.e., for establishing the existence of a mind-independent lapse of
time. Since in the quotation above Earman himself explicitly recog-
nizes that cosmic time would be a merely necessary condition for an
objective lapse of time, he cannot be interpreted as denying premise
(11) above, namely that we have independent evidence for becoming
because we have scientific evidence (as we do) for the existence of
cosmic time. Consequently, if a realist about time and becoming
wants to attack Godel's argument, she must pick up Savitt's chal-
lenge, and discuss his two premises, namely (10) and (11).
266
 On Becoming Cosmic Time and Rotating Universes

Starting with the former, could the experience of the Godelians

be identical to ours? If we grant this point, obviously we deny that
there exists a necessary link between what we experience and the
structure of objective, cosmic time also in the actual world, and it
may seem that Savitt's premise, to a certain extent, simply begs the
question. The point, however, is not that such a premise is question
begging, but rather that since the logical and physical possibility of
time travel is needed by Savitt for Godel's universe not to be ruled
out a priori, it is certainly available to an antikantian (a realist about
time) to claim that it is at least physically possible that the experience
of time of the 'Godelians' be very much unlike ours.14
Savitt might perhaps defend his premise by invoking well-known
technological difficulties entailed by time travel. Observers living in
Godel spacetime would presumably share our technological problems
concerning the amount of acceleration and fuel needed to voyage into
the past (Malament [1985]), and would not actually be travelling into
the past, though it would be physically possible for them to do so.
Moreover, we have to keep in mind that Savitt's premise (10) merely
relies on the physical possibility of their experiencing time exactly as we
do, and this point is not touched by the mere possibility of time travel.
If time travel in this context is a red herring, there is another dif-
ficulty that stands in the way of our accepting (10): to say that it
physically possible for local observers living in Godel's spacetime to
experience time as we do implies that it is physically possible for
such observers to fail to see any trace of the future. This is highly
doubtful, however, since it is certainly physically possible for them
not to be screened off from causes that are later than their effects,
exactly because they live in a universe in each point of which a
closed timelike curve can always be found! One can even argue that
in Godel spacetime there must be traces of the future, since even if the
Godelians' psychological arrow is directed along one direction of
time, and that direction is picked out as the direction of time, some
later events along that direction will have to be regarded as indi-
rectly causing events in the observers' present. So, especially if such
observers live along timelike 'loops' whose diameter is not very
large,15 we can conclude that their experience of time would be
relevantly different from ours, and Savitt's basic premise would
14
Of course, Savitt acknowledges that after a bit of scientific develop-
ment, the godelians might discover that there is no cosmic time in their
universe, i.e., no necessary structure for the existence of an objective lapse
of time [1994, p. 467].
15
This remark was raised by Joos Uffink during the discussion of the
paper.
267
Mauro Dorato

have to be abandoned. At this point, he might retort that for

observers living on very large causal loops, causes that are later than
their effects would be very improbable, and the technological diffi-
culties of travelling into the past might just make their experience
indistinguishable from ours.
Leaving to the reader the difficult task of judging who is going to
score on this uncertain point, let me strengthen my objection to
Savitt's argument by considering that also premise (11) is debatable:
is our 'direct experience of time' the only argument to believe in the
objectivity of the lapse of time? Clearly, an evaluation of this claim
depends on how to understand 'our experience of time', in particu-
lar the ambiguous and vague word 'experience'. If Savitt means to
claim that no argument in defence of an objective coming into exis-
tence is ever likely to come from physics ('experience' meant in a
very wide sense, encompassing scientific knowledge), I think we
must agree, because cosmic time cannot be regarded as sufficient for
objective becoming. Furthermore, it is certainly not among physics'
aims to yield a distinction between physical systems or entities that
are actual at a certain time and systems that are merely possible, and
precisely this distinction is needed for becoming. Consequently, in
his argument 'experience' must mean 'scientifically untutored expe-
rience'. However, even in this restricted sense, 'experience' can have
two interpretations, a broad and a narrow one.
In a broad sense, one could refer to 'experience' as it is coded in
ordinary language, particularly in those concepts—possibly a priori
for the individual but a posteriori for the species—that have been
acquired during our evolutionary history and that are tested, say, in
experiments within the so-called naive physics. These 'concepts'
(time included) must possess some sort of adaptive value, in the
sense that they must enable us to cope with the environment in a
successful way, despite their approximation and possible lack of
precision for purposes of the scientific description of the world. If
we interpret 'experience' in such a broad evolutionary, not purely
psychological, sense, we may even grant Savitt's premise (11), by
remarking at the same time that the adaptive value of our naive con-
cepts of 'object' and 'property' may justify some sort of general,
defeasible 'folk realism', telling us that such objects and their prop-
erties are prima facie real. Rather than calling into question and
'eliminate' what Sellars [1962] used to call 'the manifest image' (the
world of our experience), we may temporarily adopt its ontology,
until conflicts with the 'scientific image' force us to abandon it. On this
hypothesis, however, why doubt that there is something mind-
independent that our experience of time is about, if in our model of
268
 On Becoming Cosmic Time and Rotating Universes

the universe no scientific fact is in direct conflict with it} In our uni-
verse, unlike Gb'del's, one can be a folk realist about becoming since
a necessary condition for it—cosmic time—is indeed satisfied. In this
line of argument, rather than arguing directly for the reliability of
our experience of time, one could begin by defending, indirectly,
some sort of folk realism, which would then support in a non-ad-
hoc way also our 'natural belief in objective becoming, once con-
flicts with known physical theories are shown to be absent.
If, on the other hand, 'experience' is given a narrower, purely psy-
chological reading, isn't it quite hazardous to deny, at the present
moment, that any future philosophical arguments constructed to
prove the reality of the lapse of time must fail? For instance, Tooley
[1997] has recently given an important argument in favour of the
unreality of the future based on causation, not on our 'psychologi-
cal' experience. How can we exclude that forthcoming and more
sophisticated arguments will succeed without calling into question
our mental set-ups?
In sum, I don't mean to suggest that Savitt's reconstruction of
Godel's argument is not interesting and persuasive, but only that it
is not conclusive to establish the mind-dependence of becoming or
the ideality of time in the sense of Kant. In the remainder of the
paper, I will pick up Savitt's challenge (recall the shift of the bur-
den of proof) by defending the mind-independence of a somewhat
'deflated', minimalist and tenseless notion of becoming, which con-
cerns our experience only in the broader, non-psychological sense
mentioned above. As we are about to see, such a notion is neverthe-
less a satisfactory explication of our intuitive notion of time

5. Becoming as real occurrence of events and facts

My suggestion is to explicate, or rather simply equate becoming

with the notion of 'taking place' or 'occurring', which is also the
natural way to understand change in Broad's absolute, non-qualita-
tive sense referred to above:
Def: Becoming is real if and only if events successively and mind-
independently take place at their own proper time of occurrence.™
16
Interestingly, the etymology of 'event' betrays an original, revealing
image of motion through space, as the word comes from the Latin verb
advenire, literally 'to arrive', 'to come to', which is then extended
metaphorically to temporal matters to mean 'to occur', 'to happen', where
such happenings are changes.
269
Mauro Dorato

Given that it is non-controversial to grant that for an event to occur

at a time just means for it to exist at that time, the task that still
remains is to show that the proposed, minimalistic equivalence
between 'coming into existence at time t' (Godel's change) and
'occurring (existing) at that time' captures the essential features of
our pre-theoretical intuitions about becoming and the passage of
time.
The solution we are after is simple if we identify the lapse of time
with the view, dearest to our intuition, that the 'present coincides
with the existing'. By relativizing this claim to a time t, we get that
at t only events simultaneous with (present at) t exist, where 'exis-
tence' is here understood in a relational, tenseless sense, given by
'existence at a date/time'. Capturing this intuition in our explication
of becoming is therefore indispensable to make the latter adequate,
and it seems to me that Godel has understood this essential point
better than any other philosopher before or after him. Consider the
following, precious but strangely neglected quotation: «For that
time elapses and change exists means [...] that at any moment of
time only a certain portion of the facts composing the world exists
objectively (and different portions at different moments)* [Godel
1995, p. 235]. Provided that the notion of occurring at a certain
proper time is mind-independent—why deny that 'things occur'
and 'events happen' without our taking notice of them?—the thesis
that only the present exists (even formulated in the relativized way
seen above) is sufficient to claim that events and facts come into exis-
tence (and cease to exist) mind-independently.
In fact, how can two temporally separated events coexist in a tense-
less sense if, at any instant of (cosmic) time t, only events occurring at
t exist (at that time)? For any two temporally separated, instanta-
neous events e and /, the earlier of the two must cease to exist when
the other comes into being, provided that 'event e comes into being
(into existence) at t' (tenseless becoming) simply means 'e occurs or
happens at t' or 'e is present at (simultaneous with) t\ The first,
essential question we must face, then, is whether, and in what sense,
events can be said to coexist tenselessly in the same possible world
(spacetime), or alternatively, which arguments we have to defend
the view that only what occurs at t exists at that time. The other
problem is to show that such a relational, tenseless view of becom-
ing is a faithful explication of our experience of time and passage.
Let us examine these two issues in turn.
The argument to defend the view that at time t only what then
occurs exists as of that time may run as follows. For simplicity,
imagine a universe in which time has a discrete ordering, composed
270
 On Becoming Cosmic Time and Rotating Universes

only of instants 1 and 2, with two causally connectible, instanta-

neous events, £\ and E2, occurring at those temporally separated
instants. At 21; E2 trivially doesn't exist (both in a tenseless and in a
tensed sense of 'existence'), simply because, by definition, E2 occurs
at the different time t2\" In fact, if 'occurring at t' and 'existing at t'
must be regarded as perfectly interchangeable, tenseless expressions,
it follows that at time tx, E2 does not exist, otherwise E2 would exist
at all times (that is, in our simplified model, it would exist also at t{),
which is absurd. Therefore, since at time tj E^ exists (occurs) and E2
doesn't (in the perfectly acceptable tenseless sense seen above), one
can safely assume that E2 comes into being at t2, by simply happen-
ing or taking place at that time. Conversely, since E\ exists (occurs)
only at £1( at t2 it ceases to exist, since at that time E2 is the only exist-
ing event.
By defending such a tenseless and relational view of becoming,
one can readily join Williams [1951] in arguing that the flow of time
interpreted in a literal sense is inconsistent, because of notorious
difficulties with questions like 'how fast does the present flow'? Of
course, renouncing this view is certainly not a sacrifice, because the
explication of becoming proposed here—by broaching this problem
we come to the second issue anticipated above—does indeed save
two essential tenets of the commonsensical view of time:
(i) At any instant of time, only what is present at that time exists,
since both the past and the future at that time don't exist
(both in the tensed sense of existence, given by 'existing now'
and in a perfectly acceptable tenseless sense, given by 'exis-
tence at a date/time')18;
(ii) an absolute change in what exists can be regarded as objective,
since it coincides with the successive coming into being
(occurring) of events either (a) at different instants of a global,
cosmic time, if the latter is indeed available, or (b) at instants
of a local, proper time along a particular worldline.
I argue that such a successive coming into being of events at differ-
ent moments of time is the mind-independent, objective core lying
behind the subjective sense of literal passage of one time over another,
" For the purpose of rebutting charges of fatalism allegedly entailed by
the tenseless view of time, this point has been correctly noted already by
Oaklander [1994, 1998]. However, I think that he has not drawn its philo-
sophical consequences for the view that he himself defends about becom-
ing (he is against it).
18
The fact, urged by Savitt, that other senses of tenseless existence are
on the ground ('existing at all times' is one) is irrelevant in our context.
271
Mauro Dorato

which, admittedly, is engendered by our memory of events that

don't exist any more and our anticipations of events that are yet to
happen, fused together in a unique but continuously changing pre-
sent experience. The changing of such an experience can be
explained only by the successive coming into being of events and
states of affair at their time of occurrence.
It is in this sense that I think that such a minimalist view of
becoming, that in the literature has never been clearly formulated,
can be regarded as a tertium quid between, and therefore as a disso-
lution of, some of the main contentions between the two camps (the
'A' and the 'B') in which the analytic philosophy of time of the 20th
century has been divided. Despite the fact that tensed sentences
have tenseless truth conditions—as urged by Mellor [1981] and
Faye [1989]—becoming must be regarded, contrary to the typical B-
theorists' view, as a mind-independent feature of the universe.19 At
any instant of a cosmic or local time, tenselessly conceived events
and facts do come into being as objectively as it gets, for the simple
reasons that at any instant of time, only events occurring at that
instant exist (in the two senses seen above), and such events do not
occur all at once, but in succession. Furthermore, once we realize
that it does not make sense to ask how fast events do come into
being, because coming into being at t just means occurring at that
time and not existing before, the ghost of the infinite regress,
imported by misleading metaphors of motion through space of a
reified now, vanishes.
Incidentally, we should note that the view that mind-independent
occurring (on the part of events) is sufficient for becoming is not
completely new, since it has been implicitly defended by authors that
are usually identified as arch-enemies of becoming, like Eddington:
«events do not happen, they are just there and we come across them»
[1920, p. 51), and Weyl «the objective world simply is, it does not
happen» [1949, p. 116]. I claim that these oft-quoted passages,
whose true meaning has escaped us, are the only coherent formula-
tions of a becomingless world, i.e., a world in which events literally
don't occur, but simply are.
It should be obvious why both Weyl and Eddington defended this
view with respect to STR. Given that in this theory the temporal
order is only partial, events that are usually defined, as in Kim's
19
For a number of B-theorists defending the mind-dependence of the
difference between past, present and future, see Russell [1915], Griinbaum
[1963], Faye [1989], and Mellor [1998]. Not all B theorist defend the
mind-dependence of becoming: J. Butterfield (private communication) is
an exception.
272
 On Becoming Cosmic Time and Rotating Universes

theory [1976], by a triple constituted by a substance, a property and

a (coordinate) time, would have to be regarded as having an identi-
ty which depends on an arbitrary choice of an inertial frame. In this
case, it may appear more plausible to assume, as Eddington and
Weyl did, that events don't occur at all, but simply are, or tense-
lessly coexist in the block view of the universe. If my reading of
those oft-quoted passages is correct, these two philosophers had
already realized what I am urging here, namely that for the reality
of change and of temporal becoming, the reality of 'occurrence' suf-
fices. This, in its turn, implies that as soon as we grant that in a gen-
eral relativistic spacetime endowed with a global time order events
can objectively and mind-independently occur in succession, we
thereby introduce a change in what is real at different instants of
time for the reasons given above, and therefore a tenseless form of
becoming at a cosmic time t. It is certainly more difficult to defend
a local, worldline-dependent becoming in the Minkowskian setting,
since the present there does not extend in space but must be identi-
fied with a point (the here-now). However, this is the topic for a dif-
ferent paper.

6 Conclusion
Even if the argument above in favour of becoming were not judged
to be conclusive, what matter most for my purpose is that the adop-
tion of a relativized and tenseless notion of becoming yields a
coherent alternative to its mind-dependence and to the block view.
We can adopt the view from no-when of the block universe, a God's
eye point of view, which describes entities that are temporally
extended sub specie aeternitatis, or we can resort to a relationist, per-
spectival description of reality, which refers existence to a particu-
lar 'point of view' or instant of time. If both are compatible with
know physical theories, the choice between them can be only be a
matter of overall coherence with what else we know about the uni-
verse.
The reasons to prefer the latter view are not only pragmatic, i.e.,
given by the fact that we are temporally located beings. The former
view, by regarding the difference between present and future events
as identical to the difference between here and there, makes our
experience of time utterly unexplainable, and in principle not
describable in physicalistic or even naturalistic terms. In fact, how
can I act to produce or bring about a future event e if e coexists
(tenselessly) with the time of my action in the same sense in which
273
Mauro Dorato

a past event exists? Within the perspectival, relationist option, cau-

sation can be regarded as an ontologically loaded notion: from the
perspective of a region R, where my present action is located, events
occurring in the later region R' don't exist (tenselessly or tensedly),
and an event in R (my action) literally brings about those in R' by
causing them.
If what I am trying to argue is correct, it follows that a somewhat
deflated version of objective becoming must be reintroduced, one
that is equivalent to the notion that events mind-independently
occur at a certain proper time and place. If the proper time of a sin-
gle, fundamental particle (observer) can be extended to a cosmic
time as in standard Robertson-Walker cosmologies, becoming can
be regarded as being independent of the varying lapses of time
associated to different timelike curves, and, as such, it passes the test
of intersubjective validity, //there is a perfectly legitimate sense in
which physical events belonging to any relativistic spacetime (also
Minkowski's) exist only at their proper time and place of occur-
rence—no interpretation of relativity forces us to abandon this triv-
ially simple remark—it should be clear why this view of becoming
entails some sort of rapprochement between the so-called static view
of time and the dynamic view: the only existing facts are tenseless
(facts at times), but their becoming or coming into being at instants
of cosmic or local time is a real, though physically unexplainable fea-
ture of the universe.

References
Boltzmann, L. 1896-1898. Vorlesungen u'ber Gastheorie, transl. by S. Brush
[1964], Lectures on Gas Theory. 1896-1898. Berkeley: University of
California Press.
Broad, C. D. 1938. Examination of McTaggart's philosophy, Cambridge:
Cambridge University.
Callender, C. 1997. 'Review of M. Dorato's Time and Reality, Bologna,
Clueb 1995', British Journal for the Philosophy of Science, 48, 117-20.
Clifton, R. and Hogarth, M. 1995. 'The Definability of Becoming in
Minkowski Spacetime', Synthese, 103, 355-87.
Dieks, D. 1988. 'Discussion: Special Relativity and the Flow of Time',
Philosophy of Science, 55, 456-460.
Dorato, M. 1995. Time and Reality, Clueb, Bologna.
1996. 'On Becoming, Relativity and Non-separability', Philosophy of
Science, 64, 585-604.
2000. 'Becoming and the Arrow of Causation', Philosophy of Science,
Supplement to vol. 67, 3, S523—534.

274
 On Becoming Cosmic Time and Rotating Universes

Earman, J. 1995. Bangs, Crunches, Whimpers and Shrieks. Singularities and

Acausalities in Relativistic Spacetimes. Oxford: Oxford University Press.
Eddington, A. S. 1920. Space, Time and Gravitation. Cambridge,
Cambridge University Press.
Faye, J. 1989. The Reality of the Future, Odense, Odense University Press.
Godel, K. 1949a. 'A Remark About the Relationship Between Relativity
Theory and Idealistic Philosophy', in P. A. Schilpp (ed.), Albert
Einstein: Philosopher-Scientist, La Salle IL: Open Court, pp. 557-62,
reprinted with corrections and additions in Godel (1990).
(1949b. 'An Example of a New Type of Cosmological Solution of
Einstein's Field Equations of Gravitation', Review of Modern Physics,
21, 447-50.
(1990. Collected Works, S. Feferman et. al. (eds), Vol. 2, Oxford:
Oxford University Press.
(1995. Collected Works, S. Feferman et. al. (eds), Vol. 3, Oxford:
Oxford University Press.
Godfrey-Smith, W. 1979. 'Special Relativity and the Present',
Philosophical Studies, 36, 233-44.
Griinbaum, A. 1963. Philosophical Problems of Space and Time, New York:
A. Knopf.
Horwich, P. 1987. Asymmetries in Time, Harvard Mass.: The Mit Press.
Jeans, J. 1936. 'Man and the Universe', Sir Stewart Alley Lecture, in
Scientific Progress, New York: The McMillan Company, pp. 13-38.
Kim, J. 1976. 'Events as property exemplification', in Brand M. and
Walton D. (eds), Action Theory, Dordrecht: Reidel, pp. 159-77.
Le Poidevin, R. (ed.) (1998. Questions of Time and Tense, Oxford: Oxford
University Press.
Malament, D. 1995. 'Introductory Note for 1949b' in Godel (1995),
Collected Works, S. Feferman et. al. (eds), Vol. 3, Oxford, Oxford
University Press, pp. 261—69.
Maxwell, N. 1985. 'Are Probabilism and Special Relativity Incompatible?',
Philosophy of Science, 52, pp. 23^-3.
McTaggart, J. 1908. 'The Unreality of Time', Mind, 68, pp. 457-74.
Mellor, D. H. 1997. Real Time II, London: Routledge.
Oaklander, N., and Smith, Q. (eds) (1994. The New Theory of Time, New
Haven: Yale University Press.
Oaklander, N. 1998. Freedom and the New Theory of Time, in Le Poidevin,
R. (ed.) (1998. Questions of Time and Tense, Oxford: Oxford University
Press.
Padgett, A. 1992. God, Eternity and the Nature of Time, New York: St.
Martin's Press.
Putnam, H. 1967. 'Time and Physical Geometry', The Journal of
Philosophy, 64, pp. 240-47.
Quine, W. v. O. 1987. 'Space-time', in Quiddities, Harvard University
Press: Harvard Mass.
Rakic, N. 1997. 'Past, present and future and special relativity', British
Journal for the Philosophy of Science, 48, 257-280.
275
Mauro Dorato

Rietdijk, C. 1966. 'A rigorous proof of determinism derived from the spe-
cial theory of relativity', Philosophy of Science, 33, 341-44.
Rovelli, C. 1997. 'Half-way through the woods: Contemporary Research in
Space and Time', in J. Earman and J. Norton (eds), The Cosmos of
Science, Pittsburgh: University of Pittsburgh Press, pp. 180-224.
Russell, B. 1915. 'On the Experience of Time', Monist, 25, 212-33.
Saunders, S. 1996. 'Time, Quantum Mechanics and Tense', Synthese, 107,
19-53.
Savitt, S. 1994. 'The Replacement of Time', Australasian Journal of
Philosophy, 72, 463^74.
(2000. 'There's No Time Like the Present (in Minkowski space-
time)', in D. Howard (ed.), PSA 98, part II, Supplement to volume 67,
3, S563-S573.
Sellars, W. 1962. 'Philosophy and the Scientific Image of Man', in R.
Colodny (ed.), Frontiers of Science and Philosophy, Pittsburgh:
University of Pittsburgh Press, pp. 35-78.
Smith, Q. 2000. 'Review of Philip Turezsky's Time', The British Journal
for the Philosophy of Science, 51, 27-933.
Stein, H. 1968. 'On Einstein-Minkowski spacetime', Journal of
Philosophy, 65, 5-23.
1970. 'On the Paradoxical Time-Structure of Godel,' Philosophy of
Science, 37, 589-601.
• 1991. 'On Relativity Theory and Openness of the Future', Philosophy
of Science, 58, 147-67.
Tooley, M. 1997. Time, Tense and Causation, Oxford: Oxford University
Press.
Wang, H. 1995. 'Time in Philosophy and in Physics from Kant and
Einstein to Godel', Synthese, 102, 215-34.
Weingard, R. 1972. 'Relativity and the Reality of Past and Future Events',
British Journal for the Philosophy of Science, 23, 199-221.
Weyl, H. 1918. Raum-Zeit Materie, Berlin, J. Springer.
Williams D. 1951. 'The Myth of Passage', reprinted in R. Gale (ed.) 1967,
The Philosophy of Time, New York: Anchor Books, pp. 98-116.
Yourgrau P. 1991. The Disappearance of Time, Cambridge: Cambridge
University Press.
1999. Godel meets Einstein, New York: Open Court.

276
How Relativity Contradicts
Presentism
SIMON SAUNDERS

Introduction

But this picture of a 'block universe', composed of a timeless web

of 'world-lines' in a four-dimensional space, however strongly
suggested by the theory of relativity, is a piece of gratuitous
metaphysics. Since the concept of change, of something happen-
ing, is an inseparable component of the common-sense concept of
time and a necessary component of the scientist's view of reality,
it is quite out of the question that theoretical physics should
require us to hold the Eleatic view that nothing happens in 'the
objective world'. Here, as so often in the philosophy of science, a
useful limitation in the form of representation is mistaken for a
deficiency of the universe (Black, 1962).
The theory of relativity has excited more philosophical commen-
tary, and exerted more influence in mainstream philosophy, than
any scientific theory, with the possible exception of Newton's
theory of gravity. But it is a remarkable fact that its influence on
metaphysics proper has been somewhat marginal. That is probably
a testimony to the anti-metaphysical attitude that characterized so
much philosophy in the last century, certainly in the Anglo-
American tradition, and certainly among more scientifically-minded
philosophers. Although the hey-day of logical empiricism is long-
since past, philosophers of physics have continued to remain cool to
metaphysics. Since they remain the ones best suited to explain the
implications of relativity theory for the philosophy of time, if they
find no interesting links between these disciplines, metaphysicians
are unlikely to look for them.
I make this observation (and I promise to say no more in this
vein) because relativity theory, and specifically the special theory of
relativity, does I believe have a simple and direct bearing on a peren-
nial question in the philosophy of time. It would, I believe, settle
this question, were special relativity the whole of the story. I shall
say something about the broader perspective of quantum theory
and general relativity at the end.
277
Simon Saunders

What traditional question does special relativity decide on? It is

whether reality—what exists—is a four-dimensional web of world-
lines (a 'block universe'), or something less. There are competing
versions of what this 'something less' might be, but they are varia-
tions on what I shall call presentism, the view that only the present
is real. The argument is simple, and I will state it with the minimum
of technicality. Although, I say, it has been largely ignored, versions
of it have been stated before; it was stated in brief by Godel,1 and at
great length by Putnam2. But this latter version of the argument has
been roundly condemned by Stein. This dispute between Putnam
and Stein is in fact well-known in the philosophy of physics litera-
ture, but insofar as there is a consensus on it, it is that Stein was in
the right.3 I will come on to this dispute in due course.

Presentism

'Presentism', as I shall understand it, is the thesis that the present is

all that exists. But this needs some unpacking. It is intended to be
something more than a platitude, and surely the present is all that
exists now, as the future is all that will exist, and the past all that did
exist. Who will argue with these claims?
One way to get clearer on the presentist thesis is to say that it is
meant tenselessly; that the copula in the sentence 'the present is all
that exists' is not itself tensed. But that is a doubtful manoeuvre.
Very often the presentist will go on to deny that there is any mean-

1
For Godel the point was obvious; he immediately went on to consider
the situation in Einstein's theory of gravity, specifically in the light of his
rotating universe solutions to the field equations of general relativity. The
latter argument has been carefully analysed by a number of authors: see
Stein (1994), Earman (1996), and Savitt (1997).
2
Putnam (1967); versions of it were also stated by Rietdijk (1966) and
Maxwell (1986), and these Stein has also criticized; but their interests were
a little oblique to our topic, and their handling of it more muddled.
Rietdijk's treatment, in particular, was just careful enough, and just mud-
dled enough, to be conclusively refuted (see Torretti 1983, pp. 250—511; see
also Landsberg 1970 p. 1146-47). I shall confine myself almost entirely to
Putnam's argument, and to Stein's response to it (Stein 1978, 1991).
3
Among those who have endorsed Stein's objections to Putnam's argu-
ment, see Clifton and Hogarth (1995), Dickson (1998 pp. 165-73), and
Shimony (1993), with qualifications). So far as I know only Callender
(1998) has expressed any real reservations; but see also Saunders (1996,
1998).
278
 How Relativity Contradicts Presentism

ingful, irreducibly tenseless use of 'is', at least when it comes to the

physical world; and that passe Mellor and others, it is equally possi-
ble to give tensed truth conditions for tenseless sentences as tense-
less ones for tensed. Exercises in the philosophy of language do not
seem to be settling anything.
But nebulous though it is, there is surely something about the
presentist's position which is perfectly clear: it is intended to be a
thesis about what to count as real. It is a realist thesis. It is a claim
about temporal reality which is supposed to hold independent of
our state of knowledge or beliefs. We and our works are not what
the thesis is about; the presentist is making a claim about reality, not
about what we know or say about it.
The claim can also be put in negative terms, as the view which is
opposed to the tenseless view of time (according to which all events
exist on a par, regardless of whether we consider them as past, pre-
sent, or future). On the tenseless view, talk of events as past, present
or future is really talk about ourselves, of the relation of events to
how we are momentarily arranged. The word 'now' is like the word
'here'; mention of 'future' and 'past' is like pointing this way and
that way in space.
That is all that is needed to bring out the conflict with special
relativity. If presentism is a thesis about ontology, and says that
existence consists of a three-dimensional spatial reality; if, in
elaboration of this thesis, it opposes the tenseless view of time, and
denies that talk of events as past, present or future is elliptical talk
about the relation of events to our momentary selves: then it con-
tradicts special relativity. It contradicts it in the sense that it implies
that special relativity is badly deficient as a fundamental theory of
the world.
Of course special relativity is an empirical theory. One might take
the view that presentism is concerned with a level or reality which
is beyond the reach of experimental methods.4 But I do not grant
that physics is so limited in scope, or that metaphysics can find any-
thing deeper. But I will not argue for either of these claims here.

How Relativity Contradicts Presentism

The difficulty posed by special relativity is extremely simple.

According to presentism, all that is physically real is the present—a
4
For a metaphysician who hovers uncomfortably between this view, and
the view that anyway relativity does not contradict presentism, see Smith
(1993 pp. 2-4).
279
Simon Saunders

system of physical events all of which are simultaneous with each

other. No other events are real. Precisely what this system of events
may be, now, as I snap my fingers, may not be known to me; but
there is a fact of the matter as to what it is, and it is a universal fact
which embraces us all. It is an intersubjective reality—now, as a
snap my fingers—and it is a reality which contains us only as an
incidental part. But even if one knew all that there is to know, con-
sistent with special relativity, one would not be able to say what this
system of events might be. According to presentism, therefore, spe-
cial relativity is radically deficient as a description of reality. It is
blind to the sequencing of what is physically real.
There is no such problem in the Newtonian case. There, know-
ing all there is to know, the set of events simultaneous with this
event—as I snap my fingers—is unambiguously defined. It is all
and only those events absolutely simultaneous with it. In that
theory there is postulated a relation, absolute simultaneity, which
partitions events into disjoint classes, namely instants of time. It
does so democratically: no one event of each class is singled out in
the definition of the partition. To be precise, this relationship of
absolute simultaneity is reflexive, symmetric and transitive, so
instants of time are equivalence classes of events. This relation,
moreover, plays a crucial role in the subsequent definition of the
dynamical laws of motion (and of suitable initial data for those
laws); it could hardly play a more fundamental role in Newton's
theory. And we ourselves, and our momentary and spatial arrange-
ments, are manifestly incidental to its definition; we are incidental
to what each reality consists in. Newtonian theory, gratifyingly for
the presentist, is attentive to what is physically real. Not so special
relativity.
The argument is so simple that it speaks for itself. No technical
result is needed: it is of the essence of the theory of special relativ-
ity that absolute simultaneity as such does not exist. Everyone
knows that there is nothing else to replace it—there is no other non-
trivial symmetric and transitive relation intrinsic to Minkowski
space. Of course, making reference to the matter content of space-
time as well, there may well be methods for defining a partitioning
of spacetime into spaces (for defining global instants, as required by
presentism), but none of them are likely to claim any fundamental
status. It is unlikely that any can be taken seriously, if we are con-
cerned with the definition of the totality of what is physically real.
Only given a matter distribution of exceptional symmetry—for
example, a stream of particles all moving inertially, with zero rela-
tive velocities—would a slicing of spacetime into spaces at different
280
 How Relativity Contradicts Presentism

times (a foliation of spacetime) be obviously privileged. The pre-

sentist will literally need a river for there to be time, according to his
metaphysics.5
One need only consider the realistic candidates to see the diffi-
culty. Given any one inertial (straight) worldline, one can define a
natural slicing of Minkowski space into spaces at different times,
namely into the set of (parallel) timeslices orthogonal to it. But how
can the whole of reality—what is physically real—depend on a sin-
gle worldline? What is this thing which has this special privilege—
or who is it—that has this extraordinary status? In fact, from a phys-
ically realistic point of view, there are no objects which always move
inertially. Nor, in an infinite universe, does anyone know how to
construct such a world-line: one cannot define the centre of mass of
the universe as a whole if it is infinite; and one cannot make do with
any part of it without privileging that part. Why that part, and not
some other? Even in the finite case the option of the centre of mass
is not very attractive, involving as it does messy and arbitrary con-
ventions (there is no unique definition of the centre of mass frame
in the finite case, again in contrast to the situation in Newtonian
theory). The presentist cannot be neutral on this score; to suppose
it can be settled by convention is precisely to take the view that what
is real—the breaking down of spacetime into spaces at different
times—is not of fundamental import, but a matter of convenience,
a matter of convention. That is precisely what the tenseless theory
says. This is not an option available to the presentist.
In the general case, and in an infinite universe, in practise one
seizes on a segment of a particular worldline or worldtube of some
body (or bodies slowly moving with respect to each other). At any
point of it a spacelike hyperplane can be constructed to which it is
orthogonal. If approximately inertial, a family of hyperplanes can
be constructed which are non-intersecting, at least locally. This is
the technique used in positional astronomy: Ephemeris Time is pre-
cisely such a system for partitioning events in the history of the
solar system into spaces at different times (more precisely, it is a
based on the relative configurations of the Earth-Moon-Sun
5
For a more general criterion, less dependent on symmetry, suppose
that hypersurfaces are defined as everywhere orthogonal to the integral
curves of the four-velocity field of the fluid. Then the fluid had better be
irrotational (lacking 'twist'), if the surfaces are not to intersect. (For a
simple geometric illustration, think of a twisted rope; it cannot be cut so
as to cut each strand of it orthogonally.) Godel was led to his rotating
universe solutions to the Einstein field equations by considerations of just
this sort: see Malement (1994).
281
Simon Saunders

system). By its means, if there ever does come into being a commu-
nity of astronauts in space, moving about the solar system, it will
still be possible to agree on what events in whose lives get to take
place at the same time. The criteria will be public and intersubjec-
tive. No particular person will be singled out in counting what is
real (of what belongs to which moment in time). But the community
as a whole is singled out. The Earth-Moon-Sun system acquires a
very special status. As the basis for the criterion of what is real, it is
parochial. It will hardly do for metaphysics. It is as embarrassing as
Newton's hypothesis as to what is really at absolute rest (the centre
of mass of the Solar System). What is so special about the Earth,
Moon and Sun?
The presentist has little option but to hold out for some as yet
unknown criterion for determining what is physically real—for
splitting Minkowski spacetime into spaces at different times.
Special relativity, the presentist must conclude, is radically incom-
plete. But the alternative view is that progress in physics has
counted against presentism. Physical theories were once compatible
with it, but then they were not.

The Dispute Between Putnam and Stein

This argument is I believe unassailable, but it is similar to

Putnam's, and Putnam's has been roundly condemned. On what
grounds?
First Putnam's argument. He considers what he calls 'the view of
the man on the street'. It is the same as presentism: it is the view
that 'all (and only) things which exist now are real'. Putnam now
assumes:
I. I-now am real. (Of course, this assumption changes each
time I announce that I am making it, since 'I-now' refers to
a different instantaneous 'me'.)
II. At least one other observer is real, and it is possible for this
other observer to be in motion relative to me. (Putnam 1967 p. 240.)
He also assumes what he calls 'the principle that There Are No
Privileged Observers':
III. If is the case that all and only the things that stand in a cer-
tain relation R to me-now are real, and you-now are real,
then it is also the case that all and only the things that stand
in the relation R to you-now are real, (ibid p. 240).

282
 How Relativity Contradicts Presentism

Putnam gives no argument for this principle, and in his subsequent

use of it only transitivity is explicitly mentioned; but by the letter
of III it is clear that R must also be symmetric. And so it should be,
if of all the events which are real, no one of them is to be privileged.
If not, then if we start from a fiduciary event x (there is at least one
real event, by I), and define the others that are real as those which
stand in relation R to x, then the set {y.Rxy} that we end up with
cannot also be defined by starting with another element of this set.
In other words, each set will have to be specified by the relationship
R and a particular element of it (not any element of it); so for each
set one element of it would have to be privileged. Likewise if R is
symmetric but not transitive. Since R is surely reflexive, 'no privi-
lege' exactly forces R to be an equivalence relation. When the ele-
ments in question are events in the lives of observers—person-
stages—the principle is 'no privileged person-stages'. ('No privi-
leged observers' has a better ring.)
Finally, Putnam requires:

IV. R is definable in special relativity.

The similarity of Putnam's argument to the one I have given should
be perfectly clear. Indeed, given a partitioning of Minkowski space
M, one can always define an equivalence relation in its terms. Let
the timeslices of M—the partitions—be labelled by a parameter t,
denote {Mt}. Then R is the co-membership relation
R={<x,y>; 3t such that x e Mt and y e Mt). (1)
It is obviously an equivalence relation, and, given {Mt}, it is obvi-
ously definable in special relativity. The question that remains is
how the partitioning of Minkowski space was arrived at. If by a
relationship on M—essentially reversing the procedure just
sketched—then we are back to Putnam's approach. But whatever
the method, a principle analogous to III will apply. I would add that
not only can it not privilege any particular person, but it had better
not privilege any particular community, either. A metaphysics
which is explicitly community bound is not worthy of the name.
Putnam considered only partitions defined by a relationship R,
not the more general case. A quick result follows on the narrow
reading of IV, that R must be defined in terms of the geometry of M.
Then, trivially, there are only the two equivalence relations,
{<x,y>;x,y e M}—all events are real—and {<x,y>;x=y}—only the
fiduciary event x is real. If these are the only possibilities, it is not
hard to see which we should choose.
But Putnam did allow that R may be defined by reference to the
283
Simon Saunders

matter distribution. The result still follows, failing any special sym-
metries, but whilst intuitively plausible it is harder to prove.
Putnam did not prove it.6 He only considered the most obvious can-
didates for it, in particular Einstein synchrony, denote Ein; this
obviously fails, because for points on an arbitrary collection of time-
like lines it is neither symmetric nor transitive. He also considered
the worldline-independent relation of past causal connectibility,
denote Con. Let x < y iff x is in or on the past light cone of y; then
Con is the relation {<x,y>;x ^y}. Although reflexive and transitive,
it is not symmetric. This too is of no use to the presentist.
Putnam did not put it in quite these terms, however. In this lat-
ter part of his paper, where he introduced Con, he spoke rather of
truth-values of statements, not of the reality of events. He consid-
ered Con in this context, as the suggestion that only statements
about events in the lower half of my light-cone have a truth value.
Of this he remarked:
This last move, however, flagrantly violates the idea that there are
no Privileged Observers. Why should a statement's having or not
having a truth value depend upon the relation of the events
referred to in the statement to just one special human being, me?
(ibid. p. 246)
The point is not entirely self-evident. A statement may fail to have
a truth value because it may fail to refer to anything, and whether or
not a statement refers (and what it refers to) may well depend on
whether it has a relation to a particular human being—so much is
true of any statement containing an explicit or implicit indexical.
And Putnam did not make it clear that one can hardly insist on his
requirement I I I , of 'No Privileged Observers', as it was originally
stated, in this new context, for no-one will demand symmetry in this
case. By shifting to the question of what statements have truth-val-
ues, it is surely intended that we include statements referring to past
events as well as to present ones. Putting it in non-linguistic terms,
it is not the thesis of presentism, but rather the the thesis that only
the present and the past is real. Call it 'possibilism" If defined by a

6
It is an immediate corollary of the result of Clifton and Hogarth
(1996), namely that in the absence of symmetries, there is no worldline-
dependent relation which is so much as transitive. Perversely, they took
this to lend support to Stein's argument, rather than Putnam's (for it
generalizes Stein's proof that Con is the only intrinsically-definable non-
trivial transitive relation on M; see below).
7
In line with Savitt's terminology (Savitt 1998).
284
 How Relativity Contradicts Presentism

relation, the possibilist will obviously not want it to be symmetric.

If an earlier event is real in relation to a later one, that should not
imply that the later one is real in relation to the earlier. Given pos-
sibilism, we should not expect to obtain a democracy of timelike-
separated observers.
But possibilists will still insist on the remaining requirement built
into III, that if you are real to me, then what is real to you is real to
me as well. Transitivity is necessary here as before. And now one
might be lulled into thinking that transitivity is sufficient for possi-
bilism: that possibilism differs from presentism only in that it
demands a weaker version of III.
The waters are now seriously muddied, for of course transitivity
is not enough for possibilism. The constraint is operating, here as
before, that what is present—meaning, according to possibilism,
what is the boundary of all that is physically real—is an intersubjec-
tive and non-parochial affair. But this latter constraint can no longer
be imposed by formal conditions on R, if R relates all and only the
events which the possibilist considers as real.
Were transitivity all that is required by the possibilist, the rela-
tion Con would be just the ticket. This is exactly Stein's response
to Putnam: it is enough to define a relation 'already happened',
and it is enough if this relation is transitive and definable in terms
of the intrinsic geometry of M (reflexivity as before can be stipu-
lated). Con does the job. Stein made this clear in his first paper on
the subject (Stein 1968, p. 5); in his second he went on to prove
that Con is essentially the only such transitive relation on M (Stein
1991).
As for Putnam's error, Stein located it in the passage just cited:
The answer is that 'having or not having a truth value', in this
question, must be understood classically to mean 'at a given time'
... but 'at a given time' is not a relativistically invariant notion,
and the question of definiteness of truth value, to make sense at
all for Einstein-Minkowski space-time, has to be interpreted as
meaning 'definiteness at a given space-time point' (or event)—to
be vivid, 'definiteness for me now'. The 'Privileged Observer'
(or, rather, privileged event) is—in effect—named in the ques-
tion, and therefore has every right to be considered germane to
the answer. Putnam's objection has an exact analogue, whose
inappropriateness is plain, in the pre-relativistic case; namely, the
question 'why should a statement's having or not having a truth
value depend upon the relation of the events referred to in the
statement of just one special time, now}' (Stein 1968 p. 15).

285
Simon Saunders

According to Stein, Putnam presupposes notions that are simply

not available in special relativity. He has failed to take note of the
changed situation in that context, that 'definiteness to the present'
has to be replaced by 'definiteness at a given space-time point'.
Clearly, on making this replacement, one cannot rule out reference
to a particular point, no more than in the non-relativistic case can
one rule out reference to a particular instant.
All well and good, but clearly this changed situation is simply no
longer hospitable to presentism. In the pre-relativistic case 'an
instant' is a public reality, on which all who were included in it could
intersubjectively agree. It offered room enough for an account of
the whole of reality (it was a plausible reality, at least for some). But
nothing like this can be said of a spacetime point. If this is really all
that special relativity provides, short of the whole of Minkowski
spacetime, we have no option but to opt for the latter. If it is true
that 'the now' can only be a spacetime point—as Stein seems to
imply—then presentism is obviously untenable. A single point in
spacetime cannot be all that is physically real.
Stein is in fact perfectly indifferent to presentism.8 He makes this
clear in the paragraph that follows:
... in Einstein-Minkowski space-time an event's present is constitut-
ed by itself alone. In this theory, therefore, the present tense can
never be applied correctly to 'foreign' objects. This is at bottom a
consequence (and a fairly obvious one) of our adopting relativisti-
cally invariant language—since, as we know, there is no relativisti-
cally invariant notion of simultaneity. The appearance of paradox
only confirms that the space-time of Einstein and Minkowski is
quite different from pre-relativistic space-time, {ibid. p. 15)
The tenseless point of view is so natural that it is not even worthy
of comment: of course the present of an event is constituted by itself
alone; what else is one to think in special relativity! Stein is not con-
cerned with the metaphysical thesis of presentism. He is impatient
with talk of the view of the man in the street—he finds it curious
that special relativity should be held hostage to that—because for
Stein, it is a fairly obvious consequence of relativity theory that an
event's present is constituted by itself alone (the second of the two
trivial equivalence relations). Obviously the presentist's position is
8
Compare Torretti (1983 p. 250): 'Each event is (tenselessly) real and
determinate, in this absolute sense, as its own worldpoint. No tensed,
frame-independent statement can add to it or detract from its reality...' I
do not believe that the presentist, or anyone even loosely associated with
presentism, can agree with this statement.
286
 How Relativity Contradicts Presentism

untenable, given special relativity. Precisely so; the only question is

why Stein disputed Putnam's conclusions.
There is one more respect in which the shift of topic, to possi-
bilism and definiteness of truth value, may have led to confusion.
What is the status of future contingencies? Rietdijk advanced an
argument similar in certain respects to Putnam's, but used it to
conclude that special relativity implies determinism. Almost twen-
ty years later, and without reference to either Putnam or Rietdijk,
so too did Maxwell. It was this that prompted Stein's second
paper on the subject, in which he proved that Con was essentially
unique. It was titled 'On Relativity Theory and the Openness of
the Future'. Clifton and Hogarth's generalization of this result, to
include worldline-dependent relations (in the absence of symme-
tries), was entitled 'The Definability of Objective Becoming in
Minkowski Spacetime'. Black shifted without comment from the
view that reality is a 4-dimensional whole, to the view that change
is unreal; Shimony, approvingly citing Stein's response to
Putnam, was concerned to deny that change was illusory.
Evidently there are two further questions at stake in all this: one,
whether indeterminism is consistent with special relativity, and
two, whether the tenseless view of time is committed to the view
that change, or becoming, is unreal. But one can answer these
questions either way and yet reject presentism;9 Putnam made no
mention of either of them.
The one clear respect in which Stein flatly denies a step in the
argument as I have given it is this:
... in effect what he calls the principle of No Privileged Observers
just requires R to be an equivalence relations. But such a require-
ment has in fact no connection with the privilegedness of
observers; and it is moreover extremely inappropriate to
Einstein-Minkowski space-time—in which (unlike pre-relativis-
tic space-time, with its temporal decomposition) there are no
intrinsic geometrical partitions into equivalence classes at all, besides
the two trivial ones... {ibid. p. 19)

Stein denies that to partition a set into classes by means of a rela-

tion, in such a way that each class can be defined independent of the
9
The presentist, of course, may disagree; so too will 'A-theorists', who
find compelling McTaggart's argument that the B-series is inadequate to
the description of change. I have more sympathy for Maxwell's claim, that
special relativity poses problems for indeterminateness (see my 1996, 1998,
for further discussion).
287
Simon Saunders

choice of any particular element of that class, the relation must be

an equivalence relation. I say he is mistaken.10
How might Con be used to define a partitioning of Minkowski
space? There is a near neighbour to it which does the job quite
easily. Let x •< y iff x is on the surface of the past light-cone to y.
Define Berk as the relation {<x,y>; y •< x}. Berkeley's criterion,
indeed, was that to be real (to x) is to be seen (by x, so to be on the
past light-cone to x).u Let us now partition up M into disjoint sets
in the obvious way, as a nesting of lightcones. We do not obtain a
foliation in this way, for the partitions are not spacelike surfaces,
but put that aside. The real difficulty is that each partition has a
distinguished point—the apex of each cone—and the partitioning
as a whole clearly distinguishes a unique timelike line—the locus
of these distinguished points. Such is the price for using a relation
which is not an equivalence relation. Now suppose this partition-
ing has the metaphysical significance accorded to it by the presen-
tist: a given one of them is to define the whole of what is
physically real, a mater on which all will agree. So what is this
worldline—the worldline of what or of whom—that is to have this
extraordinary significance? It can hardly be one of the obvious
candidates from the tenseless point of view (one's own worldline,
or the worldline of the earth, or of the centre of mass of the sun,
earth and moon), wherein no ontologicai significance resides (for
on the tenseless point of view the fundamental ontologicai reality
is the whole of Minkowski space, and everything in it). We have
been over this before
On the tenseless view, all events are real; the significance of the par-
titioning is quite different. If asked which relationship gives the 'cor-
rect' partitioning, the answer will depend on how the question is con-
strued. Berk is a natural candidate, but so is Ein—in each case
referred to a particular timelike line. If the question is construed as
what to count as the past, or what to count as the present, most of us
10
Compare Sklar, '... why one should think that such a doctrine of "No
Privileged Observers" would lead one immediately to affirm the transitiv-
ity of "reality for", given that one has already relativized such previously
nonrelative doctrines as that of simultaneity, is beyond me.' (Sklar 1981 p.
130). The answer is that Putnam's aim was exactly to show that once
'simultaneity with' is relativized in a way which is non-symmetric, and
non-transitive, then 'reality for' must be relativized similarly: in which case
it is unacceptable.
11
Of course Berkeley was not proposing that all that exists is what is
visible to a unique x—a unique and particular event in space and time;
rather, he proposed that it is what is visible to God.
288
 How Relativity Contradicts Presentism

will settle for a foliation of spacetime, on which we can reach com-

munity-wide consent. It is Ein, and Ephemeris Time, that we will
choose.12

Prospects for Presentism

Putnam made mistakes in his argument, but that does not explain
its reception. Dickson has diagnosed the fault with it—as we have
seen Stein says the same—as the sheer inappropriateness, in special
relativity, of Putnam's assumptions; that 'before special relativity
can have anything to say about the doctrines in question, they must
be expressed in a language that is meaningful in a relativistic con-
text' (Dickson p. 170). Stein's sympathy with Carnap's philosophy
is well-known (Carnap was committed to the doctrine of incom-
mensurability long before Kuhn): evidently it retains its appeal for
Dickson as well.
But we can meet this objection head-on. The requirement of
inter subjectivity is certainly relativistically meaningful. The infer-
ence from that to the requirement that R be an equivalence relation
was independent of spacetime considerations altogether. T o infer
from that that there must exist a privileged foliation to Minkowski
spacetime is precisely to spell out the doctrine in relativistically
invariant terms. The fact that this doctrine is then ruled out by spe-
cial relativity was precisely Putnam's point.
At the other extreme is the deflationary, commonsensical reading
of presentism, along the lines sketched by Savitt and Moratio.
According to them, it is obvious that the past did exist, that the future
will exist, and that only the present is (presently) real. What, I won-
der, is obvious to them? That only my momentary self is (presently)
real? That what is (presently) real is what is related by Ein} Or is it
Con, or is it Berk} They may of course reply that it makes no differ-
ence which, but that is not a deflationary reading of anything.
Somewhere in between is quietism, the view that special relativity
cannot adjudicate on the matter unless supplemented by some form
or other of verificationism (Sklar 1981). So it was with Einstein's
elimination of ether; his approach to special relativity was explicit-
ly verificationist, and only given this could he conclude that the
ether did not exist. Indeed, the presentist has only to suppose that
there is, in fact, a unique resting frame, albeit that no measurement
can tell us what it is: this is the foliation he has been after all along.
12
For a recent defence of the virtues of Ein over Berk, see Sarkar and
Stachel (1999).
289
Simon Saunders

He can claim it serves an explanatory function in physics as well,

citing Lorentz, for whom it was a 'matter of taste', and citing Bell,
for whom it was the best way 'to teach the subject."3 But here I
think Stein is exactly right when he says (Stein 1991 p. 155) that if
it is verificationism that is needed, to do away with an absolute 'up'
in the face of rotational symmetry, then it is a form of it that is per-
fectly defensible, that we should all of us embrace: the form of it
which eliminates an absolute state of rest in the face of the relativi-
ty principle. And, obviously, one can draw the same conclusion on
the basis of a realist view of Minkowski space (not of course avail-
able to Einstein circa 1905).
I have maintained that presentism is a substantive position that
places clear demands on the theory of special relativity. They are
demands which I do not think can be met, consistent with that
theory. This fact needs to be clearly appreciated, if there is to be
movement on this subject. Movement there is, as soon as we con-
sider the wider perspective of Einstein's theory of gravity, and
dynamics proper.
Of course general relativity, just like the special theory, is com-
mitted to the principle of arbitrariness of foliation. Nevertheless,
for an important class of spacetime models—hyperbolically complete
spacetimes, for which the Cauchy problem is soluble—there is a nat-
ural definition of a global foliation, which has a number of desir-
able, dynamical properties. It is essentially unique; it is what is
actually used in numerical calculations in geometrodynamics; it also
has links to a number of open theoretical questions, particularly
questions concerning the nature of scale in the classical theory.
I give this example, called York time,™ after its discover James
York, not because I am convinced it is fundamental, in classical
theory, but as an example of the new avenues that are opened up as
soon as one considers gravitational dynamics proper.15 Certainly
13
This, the so-called 'Lorentz pedagogy', has recently been defended by
Brown and Pooley (2001); but in their view it is not committed to the view
that any one frame of reference is truly the resting frame; they suppose
that the forces which yield the contraction and dilation effects may be
explanatory, even if there is no fact of the matter as to what they really are.
14
For a simple exposition of its uses in Hamiltonian formulations of
general relativity, see Wald (1984).
15
And to repair my unhappy neglect of it in my (1996). The class of
hyperbolically-complete spacetimes is of course only a sector of the full
theory, and there are already problems with this in the quantum case, if
black-hole evaporation is anything to go by; but it is an important sector all
the same, and black-hole evaporation equally causes problems for unitarity.
290
 How Relativity Contradicts Presentism

none of the arguments I have given here tell against it. And ulti-
mately, of course, one must look to a quantum theory of gravity,
where the interpretation of time in canonical approaches to quanti-
zation is anyway in dispute. If one throws into the equation the
foundational problems of quantum mechanics, and the evident dif-
ficulty, in that context, of defining a Lorentz-covariant stochastic
dynamics,16 it is clear that here there is everything to play for. But
we are not about to make progress with any of these fields if the
metaphysics of presentism, in the most simple case of classical spe-
cial relativity, is still in dispute. In this most simple case, I have
argued, it can finally be laid to rest.

References

Bell, J. 1987. 'How to Teach Special Relativity', in Speakable and

Unspeakable in Quantum Mechanics, Cambridge University Press:
Cambridge.
Black, M. 1962. 'Review of G. J. Whitrow's "The Natural Philosophy of
Time"', Scientific American, CCVI, pp. 181-2.
Brown, H. and O. Pooley 2000. 'The Origin of the Spacetime Metric:
Bell's "Lorentzian Pedagogy" and its Significance in General
Relativity', in Physics Meets Philosophy at the Planck Length, C.
Callender and N. Huggett, (eds), Cambridge University Press:
Cambridge.
Callender, C. 1998. 'Shedding Light on Time', Philosophy of Science
{Proceedings), 67, S587-S599.
Clifton, R. and M. Hogarth 1995. 'The Definability of Objective
Becoming in Minkowski Spacetime', Synthese, 103, 355-87.
M. Dickson 1998. 'Digression: The Block-Universe Argument', in
Quantum Chance and Non-Locality, Cambridge University Press:
Cambridge.
Earman, J. 1996. Shrieks, Bangs, Whimpers and Grunts, Oxford University
Press: Oxford.
Godel, K. 1949. 'A Remark About the Relationship Between Relativity
Theory and Idealistic Philosophy', in Albert Einstein, Philosopher-
Scientist, A. Schilpp, (ed.), Open Court, La Salle.
Landsberg, P. T. 1970. 'Time in Statistical Physics and Special Relativity',
Studium Generate, 23, pp. 1108-59.
Malement, D. 1994. 'Commentary', in The Collected Papers of Kurt Godel,
Vol. 3, S. Fefferman, J. Dawson, W. Goldfarb, C. Parsons and R.
Solovay, (eds), Oxford University Press: Oxford.
16
See my (1996) for illustrations of the difficulties, and the reasons why
they do not arise in the Galilean-covariant case. For the difficulty in gen-
eralizating the pilot-wave theory to the relativisic case, see my (1999); the
latter are less directly linked to the foliation-arbitrariness, however.
291
Simon Saunders

Maxwell, N. 1985. 'Are Probabilism and Special Relativity Incompatible?',

Philosophy of Science, 52, 23-43.
Putnam, H. 1967. 'Time and Physical Geometry', Journal of Philosophy,
64, 240-47, reprinted in Philosophical Papers, Vol. 1, Cambridge
University Press, Cambridge, 1975.
Rietdijk, C. 1966. 'A Rigorous Proof of Determinism Derived from the
Special Theory of Relativity', Philosophy of Science, 33, 341-4.
Sarkar, S. and J. Stachel 1999. 'Did Malament Prove the Non-
Conventionality of Simultaneity in the Special Theory of Relativity?',
Philosophy of Science, 66, 208-20
Saunders, S. 1996. 'Time, Quantum Mechanics, and Tense', Synthese,
107, 19-53.
Saunders, S. 1999. 'The "Beables" of Relativistic Pilot-Wave Theory', in
From Physics to Philosophy, J. Butterfield, and C. Pagonis, (eds),
Cambridge University Press: Cambridge.
Saunders, S. 1998. 'Tense and Indeterminateness', Philosophy of Science
(Proceedings), 67, S600-611.
Savitt, S. 1994. 'The Replacement of Time', Australasian Journal of
Philosophy, 72, 463-74
Savitt, S. 1998. 'There's No Time Like the Present (in Minkowski Space-
time)', Philosophy of Science (Proceedings), 67, S563-574.
Shimony, A. 1993. 'The Transient Now', in Search for a Naturalistic World
View, Vol. 2, Cambridge University Press: Cambridge.
Sklar, L. 1981. 'Time, Reality, and Relativity', in Reduction, Time and
Reality, R. Healey, (ed.), Cambridge University Press: Cambridge.
Smith, Q. 1993. 'Introduction', in The New Theory of Time, J. Oaklander
and Q. Smith, (eds), Yale University Press: Newhaven.
Stein, H. 1968. 'On Einstein-Minkowski Space-Time', Journal of
Philosophy, 65, 5-23.
Stein, H. 1991. 'On Relativity Theory and the Openness of the Future',
Philosophy of Science, 58, 147-67.
Stein, H. 1994. 'Commentary', in The Collected Papers of Kurt Gb'del, Vol.
3, S. Fefferman, J. Dawson, W. Goldfarb, C. Parsons and R. Solovay,
(eds), Oxford University Press: Oxford.
Torretti, R. 1983. Relativity and Geometry, Pergamon Press: Oxford.
Wald, R. 1984. General Relativity, Chicago University Press: Chicago.

292
Can Physics Coherently Deny the
Reality of Time?
RICHARD HEALEY

0. Introduction

The conceptual and technical difficulties involved in creating a

quantum theory of gravity have led some physicists to question, and
even in some cases to deny, the reality of time. More surprisingly,
this denial has found a sympathetic audience among certain
philosophers of physics. What should we make of these wild ideas?
Does it even make sense to deny the reality of time? In fact physi-
cal science has been chipping away at common sense aspects of time
ever since its inception. Section 1 offers a brief survey of the demo-
lition process. Section 2 distinguishes a tempered from an extreme-
ly radical form that a denial of time might take, and argues that
extreme radicalism is empirically self-refuting. Section 3 begins an
investigation of the prospects for tempered radicalism in a timeless
theory of quantum gravity.

1. How Physics Bears on the Reality of Time

Let me begin with a quotation:

Time by itself does not exist. Time gets its meaning from the
objects: from the fact that events are in the past, or that they are
here now, or they will follow in the future. It is not possible that
anybody may measure time by itself; it may only be measured by
looking at the motion of the objects, or at their peaceful quiet.
This quote is from Lucretius's De Rerum Natura. It illustrates the
fact that, for a long time now, there have been philosophers who
have doubted the reality of time. But if that is indeed a fact, then it
seems such doubts must have been misplaced after all! Does that
mean it is simply incoherent to doubt the reality of time? I think
not. But it does mean that anyone expressing such doubts has three
tasks. The first task is to make clear just what feature of time it is
whose reality is questionable. The second task is to show how we
can get along with a concept of time that lacks that feature. The

293
Richard Healey

third task is to explain how we mistakenly came to believe in a time

with that feature.
Take Lucretius as an example. He is not denying the existence of
events, of temporal relations between them such as simultaneity,
earlier and later, of a distinction between past, present and future,
of change, or of motion. He may not even be denying the existence
of temporal congruence relations—that a definite interval of time
elapses between events. Perhaps he is merely claiming that we have
only two ways of measuring the duration of such an interval. We can
correlate its beginning and end with events in some more or less
regular motion, treated as a clock; or we can simply estimate its
duration by reference to our own internal 'sense of time'. At most
then, Lucretius is denying the existence of moments of time dis-
tinct from events that occur at them, and of an absolute temporal
metric, independent of actual physical or mental 'clocks' suited to
measure it. To make good his denial, he must show how we can
describe and explain our observations and experiences if there are
really no such temporal structures. And he owes us an account of
how we came to be fooled into believing in them.
Newton described the concept of time he was to employ in his
physics in his famous Scholium to the Principia. This was richer in
structure than that of Lucretius, incorporating not only an absolute
temporal metric, but apparently also an ontology of temporal
moments, existing independently of any events that may or may not
occur at them. Newton even endorsed the common sense idea that
time flows. But though this idea may well have had significant
heuristic value for him in developing the mathematical framework
in which to construct his theories (the calculus), it plays no essential
role in the final structure of those theories. The great predictive and
explanatory success of Newtonian physics seemed to establish the
reality of the other features of Newton's time. In retrospect this
proved a high water mark for the reality of time in physics from
which it has been receding ever since.
In the nineteenth century, Boltzmann's attempts to find a
mechanical basis for the physical irreversibility inherent in the sec-
ond 'law' of thermodynamics highlighted the temporal reversibility
of Newtonian mechanics, and indeed of all then known fundamen-
tal physical laws. This seemed to show that the distinction between
earlier and later was accidental rather than a matter of fundamental
physical law. Whether this is so remains controversial to this day.
But even if there are temporally asymmetric fundamental physical
laws, it is unclear to what extent these can account for the pervasive
temporal asymmetries we observe in physical processes. Boltzmann
294
 Can Physics Coherently Deny the Reality of Time?

went further, speculating that the manifest asymmetry between past

and future was not itself a fundamental feature of time, but rather
a reflection of the contingent asymmetries in physical processes, at
least in our region of the universe. The idea is that in so far as these
processes underlie the operation of our mental as well as physical
lives, it is ultimately this feature of our physical situation that
accounts for the perceived difference between past and future,
which is not, therefore, a real feature of time itself.
Early in the 20th century, Einstein's theories of relativity under-
mined other features of common sense as well as Newtonian time.
It came to be recognized that the temporal interval between non-
coincident events is not an invariant quantity, but depends on the
state of uniform motion to which one refers those events. If the
events happen in such a way that a material particle could travel
from one to the other, then the time interval between them is rela-
tive to the trajectory of such a particle: in that sense, time intervals
are only locally defined. If the events are space-like separated, so
not even light (in a vacuum) could travel from one to the other, then
even their time order must be relativized to a state of uniform
motion: this is the famous relativity of simultaneity. Accepting it
means acknowledging that for space-like separated events simul-
taneity, earlier and later are not two place but three place relations,
between a pair of events and a state of motion. This presents seri-
ous problems for any conception of time according to which a
single present moment separates the past from the future, since the
Minkowski space-time of special relativity (unlike that of Newton)
does not in itself determine which distant events are to count as
present.
At least Minkowski space-time permits one (somewhat arbitrarily)
to define a global present moment—most naturally as a hyperplane
of simultaneity in some chosen frame. But the space-time associat-
ed with a generic mass-energy distribution in general relativity will
contain no such hyperplanes, and may not even contain a single
global 'time-slice' (a space-like hypersurface with no boundary). As
is well known, Kurt Godel found a novel solution to the field equa-
tions of general relativity with no global time-slice and used its exis-
tence as a premise to mount a controversial argument for the unre-
ality of time. He argued that time is unreal in so far as there can be
no objective lapse of (global) time in Godel space-time. But we have
strong evidence that the space-time of our universe differs from
Godel space-time precisely in the crucial respect that it does indeed
possess a global foliation into time-slices. This not only allows for
the possibility of an objective lapse of global time in our universe.

295
Richard Healey

It also seems to guarantee that our universe changes, and indeed

expands, as time passes. For we have strong evidence that the spa-
tial geometry and matter distribution of our universe differ in just
this way on each time-slice, no matter how these slices are defined!
However, there is a very different view of how general relativity
treats time. Adopting this alternative view would void any guaran-
tee of a changing universe, and replace it with the radical denial that
there is any real physical change in a universe described by general
relativity. The alternative view has been advocated not just by some
physicists but also recently by the philosopher John Earman (forth-
coming). At first glance, it may seem merely perverse to
recommend that we adopt an alternative interpretation of general
relativity with such radical implications for the nature of time and
change. But such an interpretation can seem quite natural, or even
inevitable, from a certain perspective.
This perspective emerges from attempts to create a quantum
theory of gravity by applying standard quantization techniques to
general relativity. Such attempts have been beset for forty years or
more by severe conceptual as well as technical problems, including
the notorious 'problem of time in quantum gravity'. Here is one
way of stating that problem. Because of the vanishing of the
Hamiltonian, the quantum gravity analog to the Schrodinger equa-
tion (the Wheeler-DeWitt Equation) implies that the wave-func-
tional) that supposedly describes the evolution of space and its
contents never changes! This problem then comes back to haunt
classical general relativity. For the basic strategy behind many
attempts to quantize that theory has been to begin with a con-
strained Hamiltonian formulation, in which the theory is taken to
describe the dynamics of space and its contents, rather than as cor-
responding to a collection of models of matter distributions in
space-time. But in such a formulation, it seems that the genuine
physical quantities of classical general relativity are all constants of
the 'motion'—their values do not change! This raises two fascinat-
ing philosophical questions: Why do we experience change in such
a Parmenidean universe?' and 'Is it even coherent to suppose that an
experience of change might be an illusion}' I want to come back to
these questions after completing this initial survey of ways in which
physical theorizing bears on the purported reality of time.
As I have explained, relativity threatens the reality of various
features of Newtonian and common-sense time such as the
absoluteness of simultaneity and of temporal ordering of all events,
the absoluteness of temporal duration, and the existence of a
unique global division of events into past, present and future.
296
 Can Physics Coherently Deny the Reality of Time?

Quantum mechanics, on the other hand, seems quite conservative in

its attitude toward time. Both non-relativistic quantum mechanics
and relativistic quantum field theories simply assume some fixed
space-time background—be it that of Newton, Minkowski or a
curved general relativistic space-time. Such conservatism has even
been thought to go over into a positively reactionary attitude toward
time. I shall give three examples.
All fundamental theories known to Boltzmann were time-sym-
metric in this sense: If the models of the theory contained a motion
from state Sj at time t} to state S2 at time t2, then they also con-
tained a 'time-reversed' motion from ST2 at time tj to STj at time t2,
where ST indicates the so-called time-reversed state corresponding
to state 5 (e.g. ST might be a state of a bunch of Newtonian parti-
cles in which these particles have the same positions but oppositely
directed momenta to what they have in state S). While requiring a
slightly more subtle notion of time-reversal, relativity theory did
not affect this general feature of physical theories. But with
quantum mechanics the situation is more complicated. The time-
dependent Schrodinger equation is the fundamental dynamical
equation of non-relativistic quantum mechanics. It has been taken
to be time-reversal invariant, but this is true only if one requires the
time-reverse of the wave-function describing the state of a system
to be given by taking its complex conjugate as well as replacing t by
—t. This may he justified by claiming that the empirical content of
the theory is exhausted by transition probabilities from one state to
another, and the suggested requirements ensure that these are time-
reversal invariant. But this claim is controversial. Moreover, the
quantum measurement process seems to introduce a fundamentally
time-asymmetric element into the theory through the notorious
'collapse of the wave-packet'. Craig Callender (2000), for one, has
argued that a thorough analysis of quantum mechanics reveals that
it is not a time-reversal invariant theory. If that's right, then the
theory reintroduces the distinction between earlier and later into
physical theory at a fundamental level.
Naive versions of the collapse postulate in quantum mechanics
take it to occur at an instant, even though prior to collapse the wave
had significant amplitude over a wide region. Any such physical
process could he instantaneous in only one frame. If all collapses
occur instantaneously in a single frame, then the collapse postulate
picks out a preferred frame as a matter of physical law, in violation
of the principle of relativity. Such violation could be extremely hard
to demonstrate experimentally because of decoherence effects, and so
cannot be taken to be in conflict with existing evidence supporting
297
Richard Healey

the principle of relativity. The existence of such a preferred frame

would motivate an argument for additional temporal structure in
Minkowski space-time (or a general relativistic space-time) corre-
sponding to an absolute quasi-Newtonian time over and above the
relativistic times appropriate to (relativistic) reference frames in
various states of motion in that space-time. The presence of such
structure could be taken to reinstate absolute simultaneity and a
global distinction between the past, present and future of an event.
Similar conclusions may be drawn from a Bohmian account of vio-
lation of Bell inequalities, involving information travelling from one
wing of an Aspect-type apparatus to the other at an arbitrarily fast
speed in some unique, privileged reference frame—a frame that is
experimentally undetectable, thus preserving the principle of
relativity at an empirical level. Maudlin (1994), for one, seriously
entertains such an account.
As a third and final example of an attempt to draw a reactionary
conclusion about the real features of time from quantum mechanics,
consider John Lucas's recent appeal to quantum mechanics to locate
the flow of time at a fundamental level in physical theory. I quote:
There is a worldwide tide of actualization—collapse into eigen-
ness—constituting a preferred foliation by hyperplanes (not
necessarily flat) of co-presentness sweeping through the
universe—a tide which determines an absolute present ...
Quantum mechanics ... not only insists on the arrow being kept
in time, but distinguishes a present as the boundary between an
alterable future and an unalterable past. (Butterfield, ed. (1999),
p. 10)
Lucas believes that real quantum-mechanical collapse reinstates not
only absolute simultaneity, but also real tense, i.e. an objective but
constantly changing distinction between past, present and future
corresponding to the objective passage of events from potentiality
to actuality (or nonactuality). If he's right, then quantum physics
has finally come up with the cash to back Newton's promissory note
in his reference to the flow of time!
But this was counterfeit coinage concealed in the metaphysician's
sleeve. Even if quantum mechanical 'collapse into eigenness' were
to occur on a global time-slice this would require at most an
absolute time in the sense of a privileged foliation by such slices.
The fact that the state on a slice is not determined by those on ear-
lier slices in no way precludes the actual determinateness of states
at all time(-slice)s. An opponent who denies the metaphysical
reality of tense could even point to a sense in which 'collapse into
298
 Can Physics Coherently Deny the Reality of Time?

eigenness' renders the past more open than the future. For while a
state may collapse into any one of a discrete set of eigenstates of the
measured observable, such a post-collapse eigenstate is compatible
with each of a continuous infinity of non-orthogonal pre-collapse
states!
Quantum mechanics is a distraction from the battle over the
reality of tense, which is more properly fought on metaphysical
ground. Radical deniers of tense such as Mellor (1981) would argue
that Lucas's 'worldwide tide of actualization' falls prey to
McTaggart's (1908) notorious argument for the unreality of time.
For the kind of change Lucas takes quantum mechanics to under-
write—an event's changing from potential to actual (or counterfac-
tual) with the passage of time—is an example of what Mellor called
'McTaggart change'. The radical response to Lucas is to agree with
McTaggart that the idea of such change is simply incoherent,
though fortunately not required for things to undergo the less meta-
physically loaded kind of change we, as well as quantum physicists,
suppose them to.
If I'm right, then quantum mechanics alone neither establishes
nor poses any threat to the reality of time. But things change (if
anything does!) when one tries to come up with a quantum theory
of gravity. It turns out to be very hard to fit even ordinary change
into the resulting framework of thought. And some (notably Julian
Barbour) have given up the attempt and simply declared that
change, motion, indeed time itself are all ultimately illusory.

2. The Perils of Parmenides

Before plunging into the details of canonical quantum gravity and
the constrained Hamiltonian approach to general relativity as a
gauge theory, I want to step back to survey the ground that needs to
be covered by anyone who wishes to use these details to argue for
the unreality of time, or at least of change.
Here is the basic situation. Any theory of gravity, quantum or
classical, is a physical theory. We have no reason to believe this, or
any other physical theory, without evidence. The evidence for any
physical theory is empirical: it consists, ultimately, in the results of
observations and experiments. Whatever physical form these take,
they must give rise to experiences in scientists who perform them if
they are to serve their epistemic purpose. Such experiences will be
events—at least mental if not also physical. For there to be such
events, it must be possible to make sense of the idea that they occur
299
Richard Healey

in time—that the earlier mental state of an observer was a state of

ignorance, while his or her later mental state was a state of knowl-
edge (at least in a weak sense of that term). Moreover, at least in the
typical case, a physical theory is confirmed by testing its predic-
tions—statements made at an earlier time in ignorance of their
truth-value and then checked by making observations at a later
time. Both the formulation of a prediction and the performance of
a subsequent observation to test it are acts—events of a particular
kind involving different intentional states that the observer is in at
different times. It follows that the testing of a prediction presup-
poses the possibility of change—in the mental state of an observer,
if not also in the physical state of the world that he or she is
observing.
All these points are blindingly obvious. But note what follows
from them. There can be no reason whatever to accept any theory
of gravity—quantum or classical—which entails that there can be
no observers, or that observers can have no experiences, some
occurring later than others, or that there can be no change in the
mental states of observers, or that observers cannot perform dif-
ferent acts at different times. It follows that there can be no reason
to accept any theory of gravity—quantum or classical—which
entails that there is no time, or that there is no change. Now it is
important to note that it does not follow that no such theory can be
true. But any such theory would have the peculiar feature that, if
true, there could be no reason to accept it. To borrow a term from
Jeff Barrett (1999), any such theory would be empirically incoher-
ent. It follows that no argument that concludes that time, or at least
change, is unreal, and which starts from the assumption that some
theory of gravity—quantum or classical—is true, can have any
empirical basis. In the case of a quantum theory of gravity, this
negative conclusion may not come as a surprise. Not only do we not
currently have any convincing quantum theory of gravity, but the
prospects for finding evidence to support any such theory are at
best distant. But classical general relativity is a different matter: we
take ourselves to have considerable evidence supporting this
theory, especially following careful analysis of the binary pulsar
studied by Hulse and Taylor. But if general relativity, correctly
interpreted, implies the nonexistence of time, or of change, then
we must be wrong to take this evidence to support the theory after
all. For the correct interpretation of this supposed evidence must
undercut its epistemic credentials. Put bluntly, a radically timeless
interpretation of general relativity entails the impossibility of
performing any of the experiments and observations, the
300
 Can Physics Coherently Deny the Reality of Time?

performance of which we ordinarily take to provide our reasons to

believe that theory. Such an interpretation makes the theory
empirically self-refuting.
Now I want to suggest that things may not he entirely hopeless
for a contemporary Parmenidean. His strategy must he to embark
on an ambitious reconstruction project—the project of coming up
with serviceable replacements for those temporal concepts—implic-
it as well as explicit—which, as currently understood, presuppose
the existence of time and change at a fundamental level. A glance at
the history of science reveals a number of similar reconstruction
projects necessitated by advances in fundamental physics, some
more radical than others. As has often been noted, these have typi-
cally involved the 'demotion' of some concept, formerly assumed to
pick out some fundamental element of physical reality, to something
more anthropocentric. The up/down distinction and the distinction
between motion and rest both came to be relativized to particular
states, and so to be naturally associated with the perspective of an
observer in such a state. With relativity, the same thing happened to
spatial and temporal intervals, and also to energy and momentum.
So-called secondary qualities like colours and sounds came to be
regarded not as fundamental properties of objects and events, but
rather as corresponding to a humanly convenient way of categoriz-
ing those things in response to fundamental properties such as the
wavelengths of light and sound that they reflect or emit. There is a
tradition of describing such conceptual displacements in radical
terms. Galileo famously contributed to this tradition when he said
in The Assayer
I think, therefore that these tastes, odours, colours, etc. so far as
their objective existence is concerned, are nothing but mere
names for something which resides exclusively in our sensitive
body, so that if the perceiving creatures were removed, all of
these qualities would be annihilated and abolished from
existence. But just because we have given special names to these
qualities, different from the names we have given to the primary
and real properties, we are tempted into believing that the former
really and truly exist as well as the latter.
I place contemporary Parmenideans in the same radical tradition
as Galileo. Detecting the need for a conceptual shift in our
temporal concepts in the light of contemporary physics, they
characterize that shift in eliminativist terms. Noting the
consequent failure of our standard temporal vocabulary to mark
out any fundamental temporal facts, they take our assertions
301
Richard Healey

employing this vocabulary to be massively in error. They take the

view that the denial of time and/or change is merely the honest
acknowledgment of this error.
But even with regard to secondary qualities like colours and
sounds this radical approach is not the only way to go, and it may
not be the best way. Even if we acknowledge that the colour of an
apple is not among its most fundamental physical attributes, distin-
guishing Red Delicious from Granny Smiths by colour is extremely
convenient given the contingencies of human colour vision and
ambient lighting conditions. There continue to be good reasons for
deploying colour concepts which allow that Red Delicious apples
are red even when the lights are out, and would continue to be red
even 'if the perceiving creatures were removed'. A less radical
response to a scientifically induced conceptual shift is desirable for
practical purposes, and in this case it is certainly available. We know
it is available because we successfully avail ourselves of it on a daily
basis. But this transcendental argument from practice conceals an
important scientific and philosophical question: 'How is the human
practice of making what we call colour discriminations possible if
colour is not a fundamental property of physical objects?' Any
account of colour that denies that colour is a fundamental property
of physical objects owes us at least a sketch of an answer to this
question. And any such account that entails that the question is
unanswerable is ipso facto unacceptable.
Physical science has had the resources to provide such a sketch
since the seventeenth century. The details have been significantly
modified as the account has become more sophisticated since
Newton's classic investigations on the nature of light and colours.
But the sketch is still broadly as follows. We see a red apple when
ambient light is reflected from its surface to our eye. The surface of
the apple has intrinsic physical properties (describable without
mentioning colour) that dispose it preferentially to reflect certain
components of the ambient light incident on it while absorbing
others. The reflected light therefore has a different composition
than the ambient light: again, this composition is describable in
physical terms without mentioning colour. Light with this different
composition is disposed to elicit a characteristic sensation when it
enters the open eye of a human with normally functioning visual
and neural systems. An English speaker experiencing this sensation
has acquired the ability to apply the term 'red' to objects like apples
that elicit it in normal viewing circumstances, along with other
discriminatory and inferential abilities associated with his or her
possession of the corresponding concept. Thus while a fundamental
302
 Can Physics Coherently Deny the Reality of Time?

physical account—of apples, the light they reflect, and the human
who sees them—need include no mention of colour, it does not fol-
low that apples have no colour. Rather, that account explains our
abilities successfully to deploy colour concepts like red in ordinary
circumstances, and thereby legitimizes this application. It licenses
the claim that Red Delicious apples really are red, even though red-
ness is not a fundamental physical property.
The central claim of a contemporary Parmenidean is that time,
or something basic that presupposes time (such as change), is in
effect a secondary quality. The claim is that, like colour, time
and/or change is not a fundamental feature of the world. And just
as Galileo went on to deny the objective existence of the secondary
quality of colour, so too a contemporary radical Parmenidean
denies the objective existence of time and/or change. But while
agreeing with Galileo about the importance of the
primary/secondary quality distinction, the philosopher Locke may
be read rather as drawing a distinction between two kinds of
objective properties. On this reading, the primary qualities of an
object are those that figure in a fundamental (corpuscularian)
account of its nature: the secondary qualities arise from complex
arrangements of matter in particular circumstances that dispose it
to affect our senses in certain characteristic ways. This suggests a
moderate, neo-Lockean, alternative to the radical contemporary
Parmenidean who simply denies the reality of time and/or change.
It is to accept that time and/or change is a secondary quality, but to
go on to explain how it arises from some more fundamental
features of the world in particular circumstances that explain both
why we experience our world as temporal and why we are warranted
in so describing it. I think the contemporary Parmenidean would
be wise to take this suggestion seriously. Only in this way can he or
she rescue the physical theory that supposedly grounds
Parmenideanism from empirical incoherence.
A claim that time or change is a secondary quality requires some
initial clarification if it is to seem defensible. What is time or change
supposed to be a secondary quality of? This question reads such
claims too literally: what matters is being secondary, not being a
quality. One may explicate the claim that time is a secondary quality
as follows. Qualitative and quantitative temporal relations such as
being earlier than or occurring two weeks after are to be understood
not as external relations but as relations that hold, when they do, by
virtue of intrinsic properties of objects including their relata.
Barbour calls these objects 'Nows': a Now is something like an
instantaneous global state of the universe in relative configuration
303
Richard Healey

space. The intrinsic properties of Nows may be compared, and if all

the Nows have the right kinds of intrinsic properties, then it will
follow that certain events bear one another relations corresponding
to qualitative and quantitative temporal relations such as being ear-
lier than or occurring two weeks after. The claim then is that when
such temporal relations obtain between a pair of events, they do so
only by virtue of the nontemporal properties of all the Nows. This
is a deeply Leibnizean picture, in which neither time nor temporal
relations are literally qualities, even though both arise from what
may be considered primary qualities of Nows.
A second difficulty is presented by the inclusion of state of motion
in lists of primary qualities, beginning with Locke. For motion
clearly requires change. Indeed, Aristotle took any kind of change
in the properties of a substance to be a sort of motion, taking what
we call motion to correspond merely to change of place, or local
motion. But if motion is a primary quality, then change cannot be a
secondary quality, and nor can the time it presupposes.
This difficulty does not present a serious challenge to the thesis
that time and change are secondary qualities. To resolve it, it suf-
fices to note that the set of primary qualities should not be taken to
be defined by any fixed list. What should appear on a list of primary
qualities at any stage in the development of science are just those
properties and relations that science then takes to be fundamental.
Thus a list of the primary qualities of elementary particles today
would include such things as electric charge and intrinsic spin.
Indeed contemporary physicists have playfully added what they call
'colour' and 'flavour' to their list of primary qualities of quarks,
confident that after Galileo and Newton no-one could confuse these
with the colour and flavour of Red Delicious apples! Physics long
ago abandoned the corpuscularian restriction of primary qualities
to those observable or 'conceivable' (i.e. imaginable) in ordinary
middle-sized objects.
The fact that motion is historically taken to be a primary rather
than a secondary quality does not refute the view that time and/or
change is a secondary quality.
I hope these considerations have at least made conceptual space
for claims to the effect that change and the time it presupposes are
secondary qualities. But so far we have seen no reason why reflec-
tion on contemporary physics should motivate anyone to try to
occupy that space.

304
 Can Physics Coherently Deny the Reality of Time?

3. The Timelessness of Canonical Quantum Gravity

In the canonical quantization approach to a quantum theory of

gravity pioneered by Dirac, one begins with a formulation of gen-
eral relativity as a constrained Hamiltonian system, and proceeds to
quantize the theory by following a standard prescription that works
well when applied to other theories like electromagnetism. If one
starts with the usual variables (rather than Ashtekar's new vari-
ables), then one ends up with the Wheeler-Dewitt equation. The
equation itself is very complex, and may not even be well-defined
mathematically. I know of no realistic solutions to the full equation
(though Smolin (2001), p. 40 claims to have found some), but
approximate solutions have been found to simplified versions of the
equation at least in restricted circumstances (e.g. by Hawking and
Hartle). What would a solution look like? It would be a complex-
valued function whose arguments are 3-dimensional spatial geome-
tries with matter fields defined on them. Hartle and Hawking call a
cosmological solution a 'wave-function of the universe'. If this is
indeed analogous to an ordinary quantum mechanical wave-func-
tion, then the square of its absolute value should associate a proba-
bility to each value of its arguments. But this raises two related
problems. It is unclear what these are probabilities of: and whatever
they are probabilities of, those probabilities don't change with time
since there simply is no time parameter in the equation (it is like a
'time-dependent' Schrodinger equation with a zero Hamiltonian
operator). One might expect a solution to the equation to yield
answers to questions like 'What is the probability of finding the sys-
tem with such-and-such matter fields and spatial geometry if these
were measured at time tV But since the equation itself contains no
time parameter, any answer to such a question can only be indepen-
dent of the value of t. This would make sense if the state of the sys-
tem in fact never changed with time. One kind of Parmenidean
takes this to warrant the denial of change at a fundamental level in
any system described by the Wheeler-Dewitt equation: I shall call
this character a changeless Parmenidean. But there is an even more
radical Parmenidean who concludes not merely that any system
described by the Wheeler-Dewitt equation is in fact devoid of
change, but rather that the absence of any time parameter in the
equation shows that there is in fact no such thing as time. The idea
is that, rather than having the same answer for all values of t, a
question of the form 'What is the probability of finding the system
with such-and-such matter fields and spatial geometry if these were
measured at time tV has a false presupposition—that there are times

305
Richard Healey

to which 't' may refer! I shall call this even more extreme
Parmenidean a timeless Parmenidean. Both Parmenideans are now
committed to the kind of reconstruction project I outlined in the
previous section.
I now want to focus on the Parmenidean views of two physicists:
Carlo Rovelli and Julian Barbour. I am indebted to John Earman
(forthcoming) for his exposition of Rovelli's views, as well as to
Rovelli (1991). I take Barbour's book The End of Time and his
(1994a,b) as my source for his views. Rovelli's idea of 'evolving con-
stants' may suggest that he is a changeless Parmenidean—denying
change but not time, while it is natural to take Barbour to be a time-
less Parmenidean. But in the end I think they are both timeless
Parmenideans, though Barbour's Parmenideanism is still the more
radical.
In a constrained Hamiltonian formulation of dynamical theory, it
is normal procedure to require that genuine physical quantities be
gauge invariant, i.e. that they commute with all the (first-class) con-
straints. This requirement is motivated in part by consideration of
examples of theories such as classical electromagnetism in which
there are independent reasons to conclude that quantities that are
not gauge invariant (such as the electromagnetic potentials) are
indeed unobservable, while gauge invariant quantities (such as elec-
tromagnetic fields) are observable. In general relativity (or indeed
any diffeomorphism-invariant dynamical theory) the requirement
of gauge invariance implies that the only genuine quantities are
those that commute with the Hamiltonian constraints. But since the
Hamiltonian constraints generate the time-evolution of the system,
it follows that the only genuine physical quantities in such a theory
are constants of the 'motion'! We have a 'frozen' dynamics: no gen-
uine physical quantity changes. This looks like changeless
Parmenideanism. Note that this conclusion has been arrived at
purely at the classical level, even though one main reason to employ
the constrained Hamiltonian formulation of a dynamical theory is
as a prelude to quantizing that theory.
Rovelli's idea of 'evolving constants' can also be explained at the
classical level. The idea is that, for many constrained Hamiltonian
systems (including diffeomorphism-invariant ones), there will be
some parameter which can be thought of as a 'clock' variable—
think, perhaps, of the radius of the expanding universe in a spatially
compact model of general relativity. Now this parameter will not
itself be gauge invariant, and nor will some other quantity in whose
'evolution' one might be interested (say, a parameter corresponding
to the density of matter in that universe). But one can construct a
306
 Can Physics Coherently Deny the Reality of Time?

continuous family of gauge invariant quantities corresponding, for

example, to 'density of matter-at universe radius R', for varying R.
Now each of these is a genuine physical quantity, and while each
individual quantity is constant, the 'evolution' of the universe's
matter density may be taken to he reflected in the continuously
varying values of these quantities with varying R. We have the
illusion of change while everything really stays the same!
But how can we explain our experience of change by appeal to
such evolving constants? Earman (forthcoming) suggests two ways
to go. The first would be simply to postulate some primitive human
faculty which lets us interpret the difference between two constant
quantities such as 'matter density dj at R/ and 'matter density d2 at
R2' as an instance of change. The second, and perhaps more
promising, would be to show how the physics of the objects and our
psychology combine in such a way that we represent the world as
filled with change despite the fact that no genuine physical
quantity changes. Both these strategies strike me as hopeless. More
importantly, I don't think Rovelli himself would be tempted to
pursue either strategy. My reason for saying this is that it seems to
me that while Rovelli is indeed aware of the need to somehow
explain the temporal character of our experience, and in particular
our experience of change, he himself does not introduce his evolv-
ing constants to that end. His primary concern is to arrive at a quan-
tum theory of gravity by some canonical quantization technique.
The evolving constants are introduced not to explain our illusion of
change in a changeless world, but to provide a substitute for time in
a fundamentally timeless theory. They are there to help us to do
fundamental physics, not to explain the temporal character of our
experience. How then are we to do that? Here Rovelli appeals to a
different tactic. Two quotes are highly relevant
An accepted interpretation of [the disappearance of the time
coordinate from the Wheeler-DeWitt equation] is that physical
time has to be identified with one of the internal degrees of free-
dom of the theory itself (internal time). (1991, p. 442)
... we do not address the problem of the existence of an exact
internal time in general relativity. Instead, we assume, first, that
a way to obtain an approximate description of the world as we see
it (with time) can be extracted from the theory, second, that this
description is valid only within the approximation, (p. 443)
As I understand him, he believes that the task of accounting for the
temporal character of the world as we experience it is to be under-
taken in three stages. The first stage is to develop a coherent (and
307
Richard Healey

hopefully empirically successful!) quantum theory of gravity. The

second is to apply this to derive the existence of some kind of inter-
nal time as an approximation in a classical limit. And the final stage
is to use this approximate internal time to account for the temporal
character of the world as we experience it—since such experiences
are inevitably going to occur only under circumstances in which the
approximation of the classical limit is valid. Now Rovelli himself
says that 'The physical hypothesis that we put forward is the
absence of any well-defined concept of time at the fundamental
level', (p. 442) This makes him a timeless Parmenidean in my ter-
minology. It would be consistent with his program to develop a
quantum theory of gravity which contained nothing remotely like
time, as long as an approximate time could be retrieved from this
theory in the classical limit. But (he thinks) it turns out that the best
way of aiming for a quantum theory of gravity is to use 'evolving
constants' associated with 'clock parameters' as technical substi-
tutes for the fundamental time that is missing from that theory. I
can see nothing in the programme that requires these 'evolving con-
stants' to play any role in retrieving an approximate internal time in
the classical limit. And it is this latter project which is eventually
supposed to make the connection with our temporal experience, not
the project in which he is initially engaged in constructing a quan-
tum theory of gravity which will have this limit. Of course, we
philosophers await the successful completion of the former project
before we can be satisfied that the hoped-for quantum theory of
gravity is not merely internally consistent but also empirically suc-
cessful. For, as I stressed in the previous section, we can have no
empirical reason to believe such a theory if it cannot explain even
the possibility of our performing observations and experiments
capable of providing evidence to support it. And in the absence of
convincing evidence for such a theory we have no good reason to
deny the existence of time as a fundamental feature of reality.
Barbour explicitly denies the existence of time. Once more, his
denial is intimately connected to his attempts to make sense of the
Wheeler-DeWitt equation. Unlike Rovelli, he makes no mention of
'evolving constants', but he does make considerable efforts to show
how what is, for him, our illusory experience of time arises. His
basic explanatory device is that of what he calls a 'time-capsule'.
This is a highly-structured 'Now'. Recall that for Barbour, Nows of
various kinds and multiplicities constitute the basic furniture of the
world. To get an idea of what a Now is supposed to be, one is sup-
posed to think initially in temporal terms. In those terms, a Now
corresponds to an instantaneous relative configuration of the
308
 Can Physics Coherently Deny the Reality of Time?

universe. But of course, Nows are neither instants nor contained in

any independently existing time: they just exist atemporally. Most
Nows are not time-capsules. But amongst the vast number of Nows
are a few whose internal structure contains a representation of an
entire sequence of other Nows—a sequence that, when appropri-
ately ordered in accordance with the internal properties of each rep-
resented Now, comes to represent what looks like a possible history.
These are the time-capsules. Barbour's central idea is that experi-
encing such a time-capsule gives rise to the (misleading) belief that
it does indeed represent the sequence of events that have actually
occurred, so that the 'history' apparently represented in the time-
capsule in fact occurred as a unique sequence of events in time.
A solution to the Wheeler-DeWitt equation assigns probabilities
to all Nows, and Barbour conjectures that it must do so in such a
way as to enormously favour those that correspond to time-cap-
sules. He offers little support for this conjecture, and this lacuna has
been highlighted as the weak link in his argument by Jeremy
Butterfield (forthcoming) in his review of Barbour's book. But let's
assume that the conjecture turns out to be true. How would this
establish Barbour's claim to have accounted for our experiencing a
literally timeless world as temporal?
The truth of the conjecture would leave most of the needed
reconstruction still to be carried out. The task is to show how, in a
fundamentally timeless world composed ultimately only of Nows, it
is possible for there to be observers who naturally experience that
world as temporal—as having a history and incorporating motion
and change. To do this it would be necessary to explain why
observers in a timeless world have experiences of particular kinds,
including experiences of motion, apparent memories of past events,
observations of mutually consistent apparent records of past events,
and so on. That is what the high probability of time-capsules is
supposed to do. Barbour cannot begin to explain the character of
our experience until he has first explained how a timeless world can
contain observers capable of having experiences with any character at
all. The explanation has to start from the Nows, since Barbour takes
everything else to he composed of, or supervenient upon, these. The
first step, then, must be to provide a reconstruction of observers in
terms of Nows. Ordinarily we think of observers as enduring
embodied things that maintain their identities through time. If we
are wrong to think of observers in this way, how can we think of
them?
This is only the first of many basic questions to which Barbour
owes us an answer if he is to carry off a successful reconstruction of
309
Richard Healey

our temporal experience from timeless elements. We need to be told

what it is for an observer to have different experiences at different
times without assuming the independent existence of such times.
We need to know what it is for a particular Now to be actual rather
than merely possible. And we need to understand what the proba-
bilities generated by a solution to a timeless Wheeler-DeWitt equa-
tion are probabilities of; and how their concentration on time-cap-
sules helps to account not only for the general character of our tem-
poral experience, but also for those special experiences of scientists
capable of confirming the theory of quantum gravity on which the
whole reconstruction project rests.
In my judgment Barbour's published works do not provide clear,
consistent and satisfactory answers to these questions. But they do
contain the materials for a charitable interpretation of his project
that might do so. It is in this constructive spirit that I offer the
following answers on Barbour's behalf.
Begin with Barbour's 'many-instants' interpretation of quantum
theory, and of quantum cosmology in particular. This agrees with
Everett that there is no physical wave collapse. Barbour supposes
that the solution to a Wheeler-DeWitt equation for our universe
assumes a WKB form over a significant region of the whole space
of relative configurations, with a semi-classical factor representing
its gross features, and a quantum factor associated with the finer
details of it structure. He conjectures that the solution's probability
density is sharply peaked on time-capsules, each apparently record-
ing in its configuration a history of the development of the gross
features of the universe. Moreover, these time-capsules are sup-
posed to be arrayed along what may be called 'streamers' in relative
configuration space in such a way that those in any particular
streamer cohere with one another, in two senses. Each time-capsule
in a streamer itself contains multiple (almost) mutually consistent
apparent records of the universe's prior development. And the
capsules in a streamer may be ordered in a sequence whose elements
apparently record a single history of the gross features of the
universe up until successively later stages in its development.
Within this framework we may begin to answer the questions
posed earlier. Begin with the modal status of the Nows. The time-
capsules in each streamer collectively portray an apparent history of
a world, with relatively minor inconsistencies in their individual
representations of that history. The time sequence of a history is
portrayed by virtue of the nested time-capsules' representations:
hence temporal relations are not external relations between real
events, but relations between portrayed events determined by inter-
310
 Can Physics Coherently Deny the Reality of Time?

nal relations among the time-capsules that portray them. An event

counts as actual relative to a streamer just in case it figures in the
apparent history the streamer portrays: a Now counts as actual
relative to that streamer if and only if it contributes to the portrayal
of the apparent history. In both cases actuality is indexical. The
apparent histories portrayed in distinct streamers are like David
Lewis's possible worlds. No apparent history is any more real than
the others. In this sense all possible worlds are equally real (or unreal)!
All Nows are real. Since only those Nows within a streamer con-
tribute to a possible world, there is a sense in which only these Nows
are possible—the rest are impossible, though still real! A residual
vagueness attaches to all these categories, since the sharp peaking of
probability that defines the streamers leaves these with 'tails' of low
probability. This modal vagueness is ontological rather than con-
ceptual or linguistic, and constitutes a provocative and potentially
problematic consequence of the view.
An observer is basically a physical object whose structure permits
the formation of internal records in the configurations of its
'memory'. In the case of human observers, some of these internal
neurological record states determine the contents of experience and
conscious memory. Any enduring physical object is taken to be con-
stituted by, or at least to have its states supervenient upon, appro-
priate events portrayed by the Nows in a streamer. So now we have
our observers, and human observers with experiences. Any human
experience is determined by that human's neurological state at a
particular Now. A person will have different experiences at differ-
ent Nows. Some of these will include representations of others,
integrated in such a way as to be experienced as having happened
earlier. Others will be integrated in such a way as to be experienced
as perceived motion.
Streamers 'branch' in global configuration space. This induces
branching of the possible worlds they portray and of physical
objects, including observers, in such worlds. We suppose that the
relative multiplicities of Nows of each type reflect the probabilities
derived from a solution to the Wheeler-DeWitt equation. Consider
a physicist about to perform a quantum measurement. The physi-
cist and his environment are in a possible world portrayed by a
streamer in universal configuration space. Many different streamers
branch off from the Now that includes the physicist's experience as
he is about to perform the experiment—at least one streamer for
every possible measurement outcome. The Wheeler-DeWitt proba-
bilities give the relative numbers of streamers corresponding to
each possible outcome. The physicist's experiential state contains
311
Richard Healey

no information fixing which of these streamers portrays the 'future

history' of him and his environment. But the Wheeler-DeWitt
probabilities may be used to condition his expectations, by yielding
the probability that the streamer portraying the state he now expe-
riences also portrays this rather than that outcome of his measure-
ment. This finally gives empirical content to the Wheeler-DeWitt
probabilities. These do not specify the probability of a particular
Now being real, or actualized, or even of its being experienced as
actual. Instead, when conditionalized, they yield the probability
that an apparent history will continue in one way rather than anoth-
er. Finally we see how it is that an observer can come to have expe-
riences of a kind capable of confirming the theory of quantum
gravity that entails a Wheeler-DeWitt equation which issues in a
timeless probability density on the space of global relative configu-
rations. It is striking that experiences whose apparent content so
misleads one about the course of history may nevertheless come to
provide evidential support for a theory that predicts them. In this
respect the view involves a radical reinterpretation of the content of
experience in the tradition of Galileo's defence of Copernicanism
by reinterpreting our experience of a 'stationary' earth.
This sketch of a Barbourian account of experience is radically
incomplete and slurs over problems sufficiently serious to justify
extreme scepticism about the feasibility of the whole reconstruction
project. It is best to think of it, not as a defence of Barbour's view,
but rather as an exploration of conceptual possibilities. As such, it
serves as an illustration of the radical moves that may be required
before adopting an interpretation of a theory of quantum gravity
that demotes time to secondary quality status if one is to avoid
rendering that theory empirically incoherent.
Let me summarize my discussion of the prospects for attempts to
establish timeless Parmenideanism by reflection on the canonical
approach to quantum gravity. Rovelli acknowledges the need for a
timeless quantum gravity to explain the temporal character of the
world as we experience it, but he has a quite different motivation for
introducing his 'evolving constants'. While Barbour portrays him-
self as a radical timeless Parmenidean, he has made serious efforts
to explain what he takes to be our illusion that we inhabit a tempo-
ral world. But his efforts still fall short of what would be needed to
do that. Moreover, a successful explanation of our experience of a
temporal world on the basis of a fundamentally timeless theory can-
not establish this experience as wholly illusory. For our only
evidence for such a theory must come from experience of a
temporal world.
312
 Can Physics Coherently Deny the Reality of Time?

Now we have no current empirical evidence for any theory of

canonical quantum gravity. But we do take ourselves to have con-
siderable evidence supporting classical general relativity. This
makes it interesting to examine a recent argument by John Earman
(forthcoming) in support of an interpretation of classical general
relativity according to which there is no change in fundamental
physical magnitudes.
As I read him, Earman's argument against change in the general
theory of relativity (GTR) comes in two versions. The first begins
by noting that G T R is superficially indeterministic in the following
sense. A complete specification of the metric and matter fields on
and to the past of a time-slice S that is a Cauchy surface fixes the
development to the future of S only up to a diffeomorphism d that
reduces to the identity on and to the past of S: if mj - < M, g, T >
is a solution to the field equations, so is m2 = < M,d*g,d*T >, where
d* is the drag-along corresponding to the diffeomorphism d. The
argument continues by recommending adoption of a suggestion by
Bergman (1961) that genuine observables be restricted to diffeo-
morphism-invariant quantities. Such quantities do not discriminate
between m1 and m2, which may consequently be regarded as physi-
cally equivalent, thereby neutralizing the threat of indeterminism.
But this restriction turns out to imply the apparently absurd con-
clusion that there can be no change in local physical quantities, as
long as these are built up from Bergmannian observables.
The obvious response to this argument is to accept the absurdity
of its conclusion, and to treat it as a reductio of Bergmann's sug-
gested restriction on observables. One can still regard rrij and m2 as
equivalent representations of a single physical situation if one takes
their diffeomorphically related geometric objects (such as d*g, g) to
represent the same physical quantities. From this perspective, the
choice of d*g (say) in m2 rather than g in ml to represent the space-
time metric is fundamentally no different than the choice of one
coordinate system rather than another in which to represent the
components of g. Both are merely choices among alternative math-
ematical representations of the same physical reality. In each case,
connecting the representation to observation requires coordinating
mathematical objects to directly presented physical objects and
processes. Different choices simply require different coordinations.
The physical determinism of G T R may be secured without adopt-
ing Bergmann's suggested restriction on observables.
It is interesting that it was Bergmann who made this suggestion,
since he was an influence behind the constrained Hamiltonian
formulation of GTR, and it is to this formulation that Earman's
313
Richard Healey

second version appeals. The second version advocates a different,

though related, restriction on observables. This time, it is main-
tained that a genuine observable must be gauge invariant, in the
sense of commuting with all first-class constraints. Since the
Hamiltonian constraints generate motion, it follows that any such
observable will be a constant of the motion, again apparently imply-
ing that there is no change in any genuine physical quantity. But
despite its initial plausibility within the constrained Hamiltonian
framework, the new restriction on observables is controversial. In
particular, its applicability to the Hamiltonian constraints has been
rejected by Kuchar (1992, 1999) and others. And even its initial
plausibility depends on adopting the constrained Hamiltonian for-
mulation of GTR—A formulation that is strictly optional prior to
quantization, and only one of several frameworks in which physi-
cists have struggled to develop a quantum theory of gravity.
Another argument (not offered by Earman) appeals to the
ontology of GTR. Our common sense notion of change requires an
object that endures while having different properties at different
times. But no such enduring objects are postulated by a field theory
like GTR. Hence there can be no common-sense change in a
general-relativistic world.
This argument fails also. While GTR does not itself postulate
enduring things, neither does it exclude them. In some models of
GTR, classical fields are distributed in such a way as to provide just
the right kinds of spatio-temporal continuity to connect up what
may consequently be regarded as succeeding stages of enduring
objects that supervene on them. Moreover, GTR can be taken to
underwrite causal links between these stages even though it does not
itself make causal claims. In these circumstances, even though GTR
by itself does not entail the existence of enduring things, it makes
room for their existence. Moreover, changes in the properties of
such objects from earlier to later time-slices are naturally taken to
supervene on variations in underlying fields. Property change is not
only compatible with GTR: GTR nicely accounts for the possibility
of property change.
The standard interpretation of classical general relativity as a
space-time theory without restrictions on observables does allow for
the existence of enduring physical objects like tables, chairs, plan-
ets, stars, you and me. Moreover, this interpretation allows for the
possibility of change, including those changes in the world and our
mental states that we take to ground our reasons for confidence in
general relativity. Any interpretation that cannot allow for our
possession of this evidence is ipso facto inferior, whatever other
314
 Can Physics Coherently Deny the Reality of Time?

advantages it may seem to have. I have found wanting two argu-

ments based on alternative interpretations that involve restrictions
on what can count as observable. But if such an argument were to
succeed it could not rationally convince us to adopt a changeless-
Parmenidean interpretation of the theory absent an adequate
account of what we take to be the changes that warrant our accep-
tance of the theory in the first place. If experiments turn out to
warrant belief in general relativity only interpreted as an indeter-
ministic theory, then, surprising as it may seem, we should believe
that the world is described by an indeterministic general relativity.
The only alternative would be to suspend belief in one of our best
theories altogether. The role of the philosopher of physics as intel-
lectual conscience of the practicing physicist does not license such a
sweeping condemnation.

References

Barbour, J. 1994a. 'The timelessness of quantum gravity: 1. The evidence

from the classical theory', Classical and Quantum Gravity 11, 2853-73.
1994b. 'The timelessness of quantum gravity: 11. The appearance of
dynamics in static configurations', Classical and Quantum Gravity 11,
2875-97.
2000. The End of Time: the Next Revolution in Physics (New York:
Oxford University Press).
Barrett, J. 1999. The Quantum Mechanics of Minds and Worlds (New York:
Oxford University Press).
Bergmann, P. G. 1961. 'Observables in General Relativity', Reviews of
Modern Physics 33, 510-14.
Butterfield, J. (ed.) 1999. The Arguments of Time (New York: Oxford
University Press).
(forthcoming), 'The End of Time?', to appear in British Journal for
Philosophy of Science.
Callender, C. 2000. 'Is Time "Handed" in a Quantum World?', Proceedings
of the Aristotelian Society 121, 247-69.
Earman, J. (forthcoming), 'Thoroughly Modern McTaggart: Or What
McTaggart Would Have Said If He Had Learned the General
Theory of Relativity', to appear in Philosophers' Imprint,
http://www.philosophersimprint.org
Kuchar, K. 1992. 'Time and the Interpretation of Quantum Gravity', in
G. Kunsatter, D. Vincent and J. Williams, (eds), Proceedings of the 4th
Canadian Conference on General Relativity and Relativistic Astrophysics:
211-14. (Singapore: World Scientific).
1999. 'The Problem of Time in Quantum Geometrodynamics', in
Butterfield (ed.) (1999).

315
Richard Healey

McTaggart, J. E. M. 1908. 'The Unreality of Time', Mind 17, 457-74.

Maudlin, T. 1994. Quantum Non-Locality and Relativity. (Oxford:
Blackwell).
Mellor, D. H. 1981. Real Time (Cambridge: Cambridge University Press).
Rovelli, C. 1991. 'Time in quantum gravity: An hypothesis', Physical
Review D 43, 442-56.
Smolin, L. 2001. Three Roads to Quantum Gravity. (New York: Basic
Books).

316
Rememberances, Mementos, and
Time-Capsules
JENANN ISMAEL

Time Capsules and Presentism

I want to consider some features of the position put forward by Julian
Barbour in The End of Time1 that seem to me of particular philo-
sophical interest. At the level of generality at which I'll be concerned
with it, the view is relatively easy to describe. It can be arrived at by
thinking of time as decomposing in some natural way linearly ordered
atomic parts, 'moments', and combining an observation about the
internal structure of moments with an epistemological doctrine about
our access to the past. The epistemological doctrine, which I'll call
'Presentism', following Butterfield, is the view that our access to the
past is mediated by records, or local representations, of it. The obser-
vation is that the state of the world at any moment has the structure
of what Barbour calls a 'time capsule', which is to say that it consti-
tutes a partial record of its past, it is pregnant with interrelated mutu-
ally consistent representations of its own history.
When he speaks of time capsules, Barbour has in mind things like
tracks formed in a cloud chamber when a decaying nucleus emits an
a-particle,2 footsteps in the sand made by a passerby, or fossil
records of prehistoric animals. But to get a grip on what time cap-
sules look like in cases that approach realistic complexity, you
should think of something like Proust's Rememberance of Things
Past conceived not as an historical novel, but as a description of the
intrinsic structure of a single moment. For the book is not really
about the past, but about the traces that it has left on the present,
and what it gets exactly right is the way in which each temporal part
of one's conscious life is a kind of Barbourian time capsule. Each
living moment has written into it, into its intrinsic structure, a
representation of times that preceded, replete with their internal
representations of those that preceded them, and theirs of those that
preceded them, and so on, potentially ad infinitum.3
1
2
London: Weidenfeld and Nicholson, 1999.
3
The Mott-Heisenberg analysis of a-decay is his explicit model.
My present experience is coloured by memories, including memories
of myself remembering, and the way those memories coloured experience
under them.
317
Jenann Ismael

Barbour, as I said, isn't so literary about it, but part of the appeal of
his position is that it gets something right about the structure of a life,
and something right about our experience of time. Each moment is at
least as richly structured as Proust's novel; every moment contains
within it its very own Rememberances of Things Past. The way to pic-
ture McTaggart's B-series4 is not as a sequence of structureless points,
but a stack of novels, each as thick as Proust's, and each containing a
kind of Proustian representation of those that precede it.
Combine this insight into the internal structure of the parts of
time, with Presentism, and a gap opens up between the records and
what they are supposed to be records of (i.e., between the past and
the present representations of it)5 that Barbour exploits (and that
Bell first recommended to followers of Everett to do so) to reconcile
the appearance of historical continuity (i.e., the appearance that the
present state of the world arose as the product of continuous evolu-
tion from earlier states) with its non-actuality.6 From an epistemo-
logical perspective, the gap was always there—it was always possible
to call into question the accuracy of our historical records—(Russell
made the point with his remark that for all we know, the world was
created three seconds ago, replete with fossils, and history books,
memories, and all the rest of it)—but it is exacerbated in the context
of quantum mechanics for reasons I'll say below.
Now, go back to the stack of Proust volumes. The time capsule
structure of each means that it contains a representation of the
volumes that precede it, but there is nothing in the picture—and
nothing in particular, in the internal consistency of the various
volumes in the stack—that keeps us from stacking volumes that are
inconsistent with one another, i.e., that misrepresent earlier represen-
tations of historical events, and misrepresent them precisely as
4
'The Unreality of Time', Mind, New Series, 68, Oct. 1908. McTaggart
distinguished two series in which events were ordered, an A-series, which
ordered them in terms of their relations to the present moment, and a B-series,
which ordered in terms of their unchanging, eternal, temporal relations to one
another.
5
No restriction on the form such records take is presumed; photographs,
recordings, footprints in sand, traces in memory...
'But what is the past? Strictly, it is never anything more than we can
infer from present records. The word "record" prejudges the issue... we
might replace "records" by some more neutral expression like "struc-
tures that seem to tell a consistent story".' (Barbour, op. cit., p. 33)
6
'Quantum Mechanics for Cosmologists', in Speakable and Unspeakable
in Quantum Mechanics, Cambridge University Press, Cambridge (1987), p.
117-38.
318
 Rememberances, Mementos, and Time-Capsules

being consistent with their own depiction thereof. What we've

really got, so far as constraints on the consistency of the picture go,
is a set of novels horizontally stacked, potentially disagreeing about
historical events, but representing themselves as in accord. And the
discrepancy is undiscoverable so long as we are epistemically
trapped within the pages of a particular novel, forever creating new
records of old records, and of relations between old records and
what they are records of, but never in a position to compare any
record directly with the event it purports to record. We have no way
of establishing the reliability of the mechanisms that generate
records of past events that doesn't itself make use of those mecha-
nisms. The circle of justification is inescapable, and it is vicious.
The insight in Bell's recommendation to Many Worlds theorists
that they relinquish continuous trajectories, is that this kind of struc-
ture, internal to the parts of time, places virtually no restrictions on
the external relations between them. We can give a completely con-
sistent description of a universe constituted by a set of time capsules,
arranged in a McTaggartian B-series, containing internally consis-
tent representations of their own histories but inconsistent with one
another, and inconsistent, moreover, with the actual history, jointly
constituted by the lot of them. But Barbour takes things one step
further, and, in a bold Leibnizian move, kicks away even the exter-
nal relations between the time capsules, so that there is not, in actu-
ality, any history at all. In Barbour's picture, time capsules bear one
another internal relations of similarity and accord, but there is no
external dimension in which they are collectively ordered.7 There are
not really any genuinely external relations between time capsules,
none that don't supervene on their internal properties.
Moments, in this picture, are elements in a grand configuration
space, like the worlds in Lewis' modal universe; there is no time-like
dimension constituted by them collectively and misrepresented by
them individually. An order can be reconstructed within each
moment by stacking together internal representations of others,
using a procedure that works by comparing their intrinsic struc-
tures.8 But that, according to Barbour, is only an internal surrogate,
7
Just as, in a Leibnizian universe, there is not any genuinely external
dimension in which the monads are ordered; spatial relations, such as they
are, arise from relations among the intrinsic properties of monads. This is
why I call the move Leibnizian.
8
The procedure identifies places across time in a way that minimizes
resulting overall motion of bodies, and it turns out that both Newtonian
time, and time in General Relativity are explicitly definable in this way
from the dynamical evolution of the universe.
319
Jenann Ismael

compatible with the non-existence of any real, external time. All

that there is, on this view, are the various instantaneous configura-
tions of the universe,9 replete with their interior 'records' of other
instantaneous states, arranged in a relative configuration space by a
similarity relation expressed by Barbour's best-matching proce-
dure.10
There are a huge number of details, and they matter, but that, if
I understand it, is the broad vision. What our physics, properly
understood, gives us, according to Barbour, is a i|;-function sitting
timelessly in a relative configuration space, defining a probability
distribution that clusters on time capsules", with the appearance of
temporality arising from structure internal to the capsules. The
view is a temporal analogue of Leibnizian monadism. Whereas, for
Leibniz, space arose from purely internal relations among monads,
each of which contains an internal representation of itself as locat-
ed in a real space (i.e., in a network of external relations among spa-
tially extended systems), for Barbour, time arises from purely inter-
nal relations among monads, each of which contains a representa-
tion of itself as located in real time.
What puzzles one about the picture is the question of why, hav-
ing gone so far, Barbour stops where he does? What reason could he
have, by his own lights, for supposing the universe contains any-
thing more than a single time capsule? It's the same puzzlement one
feels, in a Many Worlds universe, about what the other worlds are
there for, or in a Leibnizian universe about why one should suppose
that there are multiple monads. The problem, in each case, is that
once you have written all worldly structure into one part of the uni-
verse, you are left with nothing for the other parts to do (except,
perhaps, if this is a proper sort of occupation—get represented, and
the thing about the other time capsules in a Barbourian universe is
that they don't even do that very well). It's as though Barbour
replaced windows with paintings of external landscapes, and then
insisted on keeping the landscapes, denying even, that they were
faithfully depicted.12
9
All those ascribed a non-zero amplitude by the quantum state.
10
E.g.: is time atomic? How big is the smallest time-capsule? Endless
technical details, and questions of motivation.
11
Barbour is up-front about the conjectural nature of the assumption
that the i|/-function will end up clustering on time-capsules; the position
could fall on this purely technical consideration.
12
At least in the case of monads, their internal structure really does
reflect the actual network of relations, though it turns out that they are
internal.
320
 Rememberances, Mementos, and Time-Capsules

Suppose we parted ways with Barbour on this question and held

that the universe consists of a solitary Barbourian time capsule: a
single temporal monad. It would be hard to say, in that case, what
remains of Barbour's denial of the reality of time. It is an essential
part of the view, crucial to its empirical plausibility, that time cap-
sules have an internal surrogate for time (i.e., parts that represent
parts of time, and that can be ordered by a best-matching procedure
into a representation of history), and the question is, what is miss-
ing in a universe that consists of a single capsule, to make one want
to describe it as a time-less one? What distinguishes it from an ordi-
nary McTaggartian B-series? And if nothing, why would one
describe a view that holds that the universe consists in a whole big
bunch of these as one that denies the existence of time? Isn't it
rather a temporally rich universe? What this puts pressure on is the
very difficult analytic question, raised by any view that denies the
reality of time, of what it is, exactly, for time to exist.
I can think of a couple of things that might make one resist
describing the single capsule universe as an ordinary temporal reali-
ty. One is that while a time capsule has an internal time, it is not itself
extended in time. It is like a book with parts that represent parts of
time, but that are not themselves arranged in it. The second is that
the gappiness of historical records in a time capsule universe has to
be given an ontological interpretation. The history of such a uni-
verse has to be as spotty as our records of it; if there is nothing in its
occurrent state to determine the precise moment between two times,
t and t*, that an a-particle is emitted from a radioactive atom, then
there was no such moment (although it will still be correct to say that
the particle was emitted, and indeed emitted between t and t*).13
Both considerations seem too esoteric to underwrite the denial of
the existence of something that plays such a central role in so much
of our thinking about the physical world. And Barbour, in any case,
holds firmly to the existence of all time capsules assigned a non-
zero probability by the i|/-function. His reason is this:
'I believe all of Platonia is "there", not just a single time capsule,
because there is then at least some chance of explaining why I
experience this instant (because it is one of many to which the
wave function of the universe gives a high probability). So the
mere fact that I experience this instant with properties that (in
13
Suppose that t is the time at which we prepared the particle, that t* is
the instant, 5 seconds later, at which it is first detected outside the nucleus,
and that there is nothing to place the emission event at any moment
between t and t*.
321
Jenann Ismael

principle if my scheme is correct) theory predicts allows me to

conclude that the others must be experienced too.'14
I'm not sure what to make of this, but I'll let Barbour have the last
word, and turn to a couple of features of the general metaphysical
picture that make it interesting from a philosophical perspective.

Temporal Leibnizianism

The first is something I have already noted: the Leibnizian struc-

ture of the Barbourian universe. Leibniz' and Barbour's pictures
both make something of the fact, which is both a consequence of
our physical theories, and a salient feature of our experience of the
world, that every part of space and time has written into its mater-
ial contents—i.e., into the structure of the concrete bodies, and the
waves it contains, into the memories of the people that occupy it,
and the books and sounds they produce—variously complete and
variously faithful representations of other parts.
Our physical theories provide us with increasingly detailed
accounts of natural mechanisms that give rise to this kind of struc-
ture (theories of wave-propagation that tell us how waves carry
structure from one part of space to another; neurophysiological the-
ories that tell us how world-representing structure gets built into
the wet stuff between our ears, theories we can add to our practical
understanding of how to build the structure in our heads into our
material surroundings). It is something that Bohm calls attention
to repeatedly, and that plays a central role in his own philosophy;
'consider ... how on looking at the night sky, we are able to dis-
cern structures covering immense stretches of space and time,
which are in some sense contained in the movements of light in
the tiny space encompassed by the eye (and also how instruments,
such as optical and radio telescopes, can discern more and more
of this totality, contained in each region of space)." 5
And rather than reify the global structure in each of the parts,
Barbour and Leibniz both deny the existence of anything over and
above the parts, and think of temporality and spatiality, respectively,16
as emerging from internal relations among, or structure internal to,
14
Personal communication.
15
'Quantum Theory as an Indication of a New Order' in Wholeness and
the Implicate Order, Routledge, New York (1980), p. 149.
16
Or, in Barbour's case, the appearance of it.
322
 Rememberances, Mementos, and Time-Capsules

the parts. Barbour does with time essentially what Leibniz does
with space, and what Bohm does with space-time. There are differ-
ences, to be sure: Barbour takes time as the basic structure, denies
the existence of genuine temporal relations between time-capsules,
and doesn't place restrictions on the internal relations among them.
Leibniz takes space as the basic structure, and denies the existence
of spatial relations (conceived non-reductively as genuinely external
relations between monads), but places strong consistency con-
straints on internal relations among them.1718 Bohm takes space-
time as the basic structure, conceives of spatio-temporal relations as
full-fledged, external relations between point-like events, and has
much that is new, and very interesting, to say about the particular
way in which global structure is locally represented. But all of them
have in common the basic construction of a whole constituted by
parts that contain (variously complete and variously accurate) rep-
resentations of it.

Memories and mementos

The second thing I want to consider is the relationship between
Barbour's metaphysical view and quantum mechanics. The insight
Barbour took from Bell was that one can deny that the actual history
of the world is continuous, while explaining the appearance of
continuity by pointing to quantum mechanisms for the creation of
consistent records. The idea was that once you've got all this past-
representing structure written into the present, acknowledged that
knowledge of the past is mediated by knowledge of present past-
representing structures, and accepted a quantum-mechanical
account of how such structures (i.e., records) are produced, you
realize that the appearance of consistency (which is to say, a positive
result for measurements to check the accuracy of our records of the
past) places virtually no constraints on the actual relations between
those records and the past, or even, indeed, among those records
themselves. The history of the world may jump around as discon-
tinuously as you please from one moment to the next, without any
threat to the appearance of continuity, and without any way of dis-
17
The internal structures of monads have to be unfolding in sync with
one another in a way that gives rise to the impression of a common exter-
nal cause.
18
The other difference, of course, is that Barbour describes his view as
a denial of the reality of time, where Leibniz describes his as a view about
the true nature of space.
323
Jenann Ismael

covering that the physical mechanisms that produce records are

anything but what they purport to be: more or less reliable ways of
generating faithful representations of past events. Barbour did Bell
one better by denying that there are trajectories, continuous or oth-
erwise. All that exists, according to him, is the collection of time
capsules, and the ijj-function giving their relative probabilities.
Butterfield, in a review of Barbour's book,19 points out that
Presentism, as a philosophical doctrine, isn't inevitable. And that is
correct. But it does express the epistemological position in which
our physical theories (both classical and quantum) cast us, and that
gives it something more than the status of an optional, and not espe-
cially attractive, philosophical view. We simply cannot accept a
broadly naturalistic picture of ourselves and deny that our knowl-
edge of distant places and times is mediated by local representa-
tions. He also remarks that Barbour's position makes as much sense
in classical contexts as in quantum mechanics, and that too is cor-
rect, but it leaves out Bell's central insight.20 It is true in the classi-
cal world, as surely as the quantum one, that we are trapped in the
present, forever planning new measurements to check results of old
ones, but no more able to check whether the new are consistent with
the old than whether the old are consistent with what they measure,
i.e., forever creating records of past events, and records of the rela-
tions between those records and the events they record, with no
independent way of ascertaining whether the mechanisms for creat-
ing records really do that, no way of telling for sure, that is, whether
they actually reveal pre-existing consistency.21
The twist added by quantum mechanics is that it elevates what
was in the classical case a mere epistemic possibility (viz., that
19
Butterfield, BjfPS, forthcoming. The article surveys the whole of
Barbour's work and situates it with respect to the contemporary philo-
sophical literature about time.
20
Bell himself dismisses the view on the grounds that it gives rise to
scepticism:
"Everett's replacement of the past by memories is a radical solipsism -
extending to the temporal dimension the replacement of everything
outside my head by my impressions, of ordinary solipsism or posi-
tivism. Solipsism cannot be refuted. But if such a theory were taken
seriously it would hardly be possible to take anything else seriously."
(Bell, op.cit.,p. 136).
21
There are two ways to think of the relationship between records and
measurements; you can think of measurements as interactions that create
accessible records of not otherwise accessible facts, or you can think of
records as the presently accessible results of measurements on the past.
324
 Rememberances, Mementos, and Time-Capsules

measurements don't simply create records of preexisting facts) to a

nomological necessity. The theory actually entails non-canonical
relations between measurement results and the events they are sup-
pose to record.22 Whereas in classical mechanics, the physical laws
entail that a photograph, a footprint in the sand, or in general, the
position of a pointer observable after measurement can only have
arisen by a deterministic process from the preceding events of
which they constitute records, in quantum mechanics, the physical
laws themselves block any direct backward inference from the result
of a measurement to the state of the world beforehand. Whereas in
classical mechanics, one cannot deny the faithfulness of records and
the reliability of the processes that generate them without denying
the physical laws, the laws of quantum mechanics themselves entail
that records aren't generally reliable. For recording is just a kind of
measuring, one that takes the present observable state of the world
as a pointer observable in a measurement of its state in the past.
And we know that we cannot in general interpret the results of indi-
vidual quantum mechanical measurements as simple, faithful repre-
sentations of the state of the measured system.
There is a film in theatres now called Memento in which the hero
doesn't have a short term memory, and has to rely for his informa-
tion about the recent past, on various kinds of material artifacts:
snapshots, written notes, tattoos, what other people tell him. We
think that the fact that we have memories puts us in a better posi-
tion, but if memories are just tattoos in the brain—i.e. present rep-
resentations of the past events—the difference is shallow. They are
only as reliable as the processes that produce them.
Conceived naturalistically, memory has the same status as per-
ception; both are physical processes that generate local (respectively,
present/internal) representations of distant (past/external) states of
affairs.23 Doubt is appropriate if there are occasional, contextual
reasons for thinking that the mechanisms aren't functioning
normally (things are broken, conditions are non-standard, or some
22
The only thing we can conclude from the result of an individual mea-
surement is that the measured system is not (or, is with measure zero prob-
ability) in an eigenstate of the measured observable with eigenvalue
orthogonal to the one observed.
" There are philosophical positions (sometimes called 'direct realist')
that hold that both perception and memory, are representationally
unmediated ways of apprehending external things and the past. So long,
however, as we are spatially localized things picking up information about
our environments from local causal interactions, Presentism is the episte-
mology built into our physical theories.
325
Jenann Ismael

such), but global scepticism is possible only by denying that the

mechanisms operate as they are supposed to, i.e., by holding a spe-
cial, non-standard theory about their operation. That is what tradi-
tional sceptical possibilities offer; whatever else it is, Descartes' evil
daemon is an alternative hypothesis about the mechanisms that give
rise to our perceptual states.
The lesson that Barbour took from Bell, and that we can learn
from him, is that quantum mechanics provides just the kind of non-
classical account of the generation of records that undermines their
general reliability as sources of information about the past. The
insight is that in a quantum context, the mechanisms that generate
future representations of moments past leave us with something
that bears as loose a relation to their source as the post-measure-
ment position of a pointer observable to the pre-measurement value
of the measured observable. The hero of Memento gets into all sorts
of difficulties because his records are generated by unreliable mech-
anisms (they are produced by people [himself included] and hence
dependent for their reliability on the trustworthiness of their pro-
ducers). If we call records like that 'mementos', we can put the les-
son by saying that, where classical physics gave us memories, quan-
tum mechanics gives us only mementos.24

Queries

A couple of final questions. There is an irony in the fact that at just

the point that Barbour thinks physics has divested itself of all
vestiges of temporality, if what I have been suggesting is correct, it
actually does a fair job of capturing central features of the experi-
ence of time. One of the most surprising things about Barbour's
view is just how much of our temporal experience can be recovered
from structure internal to his time capsules.25 This raises again the
question of why Barbour describes his view as a denial of the reality
of time, rather than an idiosyncratic theory about what time is. The
question is not inconsequential. It is, of course, in its general form
('What does the world have to be like for time to be real?', or, more
24
Or, a formulation I prefer, the classical world remembers its earlier
states, where the present state of the quantum universe is merely a memen-
to of its past.
25
I have argued, independently that we can even find in the relations
among temporally situated representations of time—of which time cap-
sules are instances—something that satisfies McTaggart's desiderata for
passage ('The Reality of Time', ms.).
326
 Rememberances, Mementos, and Time-Capsules

pointedly, 'What properties must a physical parameter have to

deserve to be called "time"?') the question that confronts the quan-
tum gravity theorist.
Another question that it would be good to hear more from
Barbour about has to do with records. Time capsules are records of
a certain kind, specifically, structures that encode an appearance of
history. But he doesn't give any explicit, intrinsic characterization of
what this means. In the classical contexts, we could think of them as
structures generated by mechanisms that lend them a kind of nat-
ural intentionality; we could say that A is a record of B iff B was
caused in the right way by A, or, perhaps, if B was typically a reli-
able sign of a preceding A-occurrence, or if A gave rise to B by a
structure-preserving process, or some such thing.26 The problem is
that in a quantum context, reliable covariation is out of the picture,
and Barbour forswears external connections, so causal relations, at
least as usually conceived, aren't available to him. In virtue of what,
by Barbour's lights, does an instantaneous configuration (e.g., a
footprint in the sand, a track in a cloud chamber) constitute a record
of this or that sort of preceding event.
Questions about what it is for a structure to have representation-
al purport, and to have the particular purport that it does, are noto-
riously hard; the reason it is fair to demand something more from
Barbour in the way of an explicit account is that his central notion
is ill-defined without one, and none of the approaches in the litera-
ture would seem to serve his purpose.27 There are some indications
in his discussion of the Mott-Heisenberg analysis of a-decay of
how it might go, but one would like to see it worked out.
There is a very great deal more to say about the view; I have
focused on features that are especially suggestive from a philosoph-
ical perspective: in particular, the surprising degree to which our
experience of time can be recovered from structure internal to its
parts, and the insight about the looseness, in a quantum context, of
the relationship between the past and our present representations of
it. There are other aspects of the position, also of philosophical
interest, and a number of deeply perplexing issues, that I haven't
26
There are a variety of accounts in the literature, all presupposing some
form of causal determination or nomological covariation.
27
The derived intentionality of artifacts like linguistic structures,
designed with representational intent isn't obviously applicable (unless
the intent is God's, and Barbour wants to convict him of malice). One
might surmise, however, from some of his remarks about consciousness,
that Barbour inclines towards some sort of irreducible intentionality
derived from their relations to human minds.
327
Jenann Ismael

touched on (not to mention endless questions, of both a technical

and conceptual nature, concerning the physics and the relationship
to quantum gravity).28 It is bound to take some time before the view
is fully absorbed, and I am not sure I have understood it entirely,
but it seems to me a genuinely new position, with deep and acknowl-
edged affinities to Leibnizian monadism, that is bound to repay
philosophical attention.

28
There are a whole set of questions, for instance, about what Barbour
means when he talks about selves; he speaks sometimes as though he is a
self-aware time-capsule, and sometimes as though he thinks he is tempo-
rally extended, 'present', somehow, in different time capsules.
328