JOHN W.

CARROLL

INSTANTANEOUS MOTION

(Received in revised version 26 April 2002)

ABSTRACT. There is a longstanding definition of instantaneous velocity. It says

that the velocity at t0 of an object moving along a coordinate line is r if and
only if the value of the first derivative of the object’s position function at t0 is r.
The goal of this paper is to determine to what extent this definition successfully
underpins a standard account of motion at an instant. Counterexamples proposed
by Michael Tooley (1988) and also by John Bigelow and Robert Pargetter (1990)
are reinforced and illuminated by considering the presence or absence of changes
to the object’s motion.

Our everyday thought and talk are strewed with mentions of objects’
having properties at an instant. For example, a few minutes ago, I
told a student that I would be in my office at 1:30 P.M. today. My
statement was consistent with my not being there at 1:29:59 P.M.
and consistent with my not being there at 1:30:01 P.M. In fact, my
statement was consistent with my not being there 0.0 . . . 01 seconds
before or after 1:30, no matter how many zeros are filled in.1
There is some resistance among philosophers to taking, at face-
value, apparent thought and talk of an object’s having a property at
an instant. Thanks to Zeno, this is especially true when the property
in question is motion. Sometimes, seeming allusions to motion at an
instant are taken to be descriptions of motion over a small temporal
interval (cf., “For at a moment it is not possible for anything to be
either in motion or at rest”, Aristotle, Physics 239b.) More often,
motion at an instant is taken somewhat more seriously; it is thought
to be definable in terms of motion over intervals. On this view,
instantaneous velocity is defined in terms of the object’s interval
velocities (cf., Anton, 1992, p. 311):
If s(t) is the position function of an object along a coordinate line, then the object’s
ds
velocity at time t0 is defined by: v(t0 ) = dt 0
= limt →t0 s(t t)−s(t
−t0 .
0)

(‘ds/dt’ is the symbolic name for the derivative of the function s(t),
which is defined by the given limit expression.) What this definition

Philosophical Studies 110: 49–67, 2002.

© 2002 Kluwer Academic Publishers. Printed in the Netherlands.
50 JOHN W. CARROLL

says about the motion of an object along a coordinate line is that

the velocity of the object at t0 is r if and only if the value of the first
derivative of the object’s position function at t0 is r. Basically, we are
to look at the slope of the line formed by the point on the graph at t =
t0 and other points on the graph to its left and right, looking at points
that are closer and closer to the point on the graph at t = t0 . If those
slopes get and stay arbitrarily close to some value, then that limit is
the velocity of the object at t0 . If those slopes do not get and stay
arbitrarily close to some value, then that limit doesn’t exist and so
neither does the object’s velocity. For many philosophers, instantan-
eous motion is nothing more than having a nonzero velocity as just
defined, instantaneous rest nothing more than having such a zero
velocity. Call this the Russellian view (cf., Russell, 1903, p. 473).
Call velocity as defined in the standard definition above, Russellian
velocity.2
In this paper, I will argue that the Russellian view does not
adequately characterize our common-sense concept of motion at
an instant, the one that is part of our everyday thought and talk.
My conclusion is in line with the conclusions advanced by Michael
Tooley (1988) and also by John Bigelow and Robert Pargetter
(1990). In fact, some of their examples will provide a starting point
for my paper. Previously, there has been almost no detailed discus-
sion of their arguments. This is surprising because, despite the
lack of discussion, the issue has been recognized as an important
one in contemporary metaphysics: Some 4-dimensionalists and
other philosophers3 concerned with the viability of the doctrine
of temporal parts have shown a great deal of interest in a non-
Russellian concept of velocity. This interest stems from a concern to
maintain some doctrine of Humean supervenience (along with the
doctrine of temporal parts) in the light of Armstrong/Kripke-style
homogeneous-sphere counterexamples. My suspicion is that part of
the reason for the sparse attention is that, as they are presented by
Tooley and by Bigelow and Pargetter, the examples call for intuitive
judgments about the motion of an object at the particular instant
in question that many philosophers are not comfortable making.
What I plan to do is to consider some of their cases from a new
perspective. The shift that will provide the novel perspective is this:
I will describe hypothetical situations similar to those described by
 INSTANTANEOUS MOTION 51

Tooley and by Bigelow and Pargetter. They all involve an object that
exists for a nonzero temporal interval. Then I will ask the reader to
make a judgment of whether there was a change in the motion of the
object during that interval or ask the reader to make a counterfactual
judgment of whether there would have been a change in the motion
during that interval if certain events had not occurred. We are often
prepared to make these judgments without making any prior judg-
ments about the motion of the object at the particular instant in
question. These change-of-motion intuitions will reinforce the prob-
lems for the Russellian view revealed by Tooley and by Bigelow and
Pargetter.
This is a project that is squarely within contemporary meta-
physics. My project is analogous to metaphysicians’ attempts to
criticize some extant account of, say, lawhood or causation. I will
show that the Russellian view does not state necessary and suffi-
cient conditions for instantaneous motion by presenting possible
situations where it is plausible to think that this view is false. As is
the norm with critical metaphysics, the issue is whether the view in
question is necessarily true. My arguments do not clearly undermine
the truth of the Russellian view and hence do not clearly undermine
the uses physics makes of the standard definition of velocity. Indeed,
as I speculate in the final section of my paper, I believe that this
is one reason why the Russellian view remains the received view.
This does nothing to diminish the importance of my thesis. Aside
from my thesis’s straightforward importance to better understanding
the nature of motion, it bears on how exactly physical theory might
explain the motion of objects.

1. NO CHANGE, NO DIFFERENCE

Suppose an object, b, is traveling in a straight line at a constant

velocity in pretty much empty Newtonian space. It may be experi-
encing gravitational forces from distant planets but these all balance
out. b’s position function is given by j(t) = 2t if t ≤ 2. The function is
undefined for t > 2 because t = 2 is the last moment of b’s existence.4
What happened? Well, let us suppose that a powerful machine akin
to Sydney Shoemaker’s (1984, p. 241) table canceller is fired at b
at t = 2. The activation of this machine and b’s nonexistence are not
52 JOHN W. CARROLL

simultaneous; b is still in existence at t = 2. But the effects of the

machine are immediate: At all times after t = 2, b is not in existence.
One instant it is there, intact, and unaffected. At all later instants,
there is nothing left of the object.
b has no Russellian velocity at time t = 2. It is not that it has a
Russellian velocity of zero. It is that the Russellian characterization
of velocity leaves b’s velocity undefined. Now, what is interesting
about this example is that, though b does not have a Russellian
velocity at t = 2, it is tempting to think that it is moving at t = 2.
One way of motivating this common-sense judgment plays off
intuitions about the intrinsicness of motion. Consider b in compar-
ison to another object similar to b in every respect except that it does
not go out of existence. Over the closed interval from t = 1.5 to t =
2, the motions of these two objects are perfectly similar. The only
significant differences in their motion appear after t = 2 when b is
out of existence. But, it is hard to see how these differences could
have consequences for what the velocities of those objects are at t =
2 (cf., Tooley, 1988, p. 242).
There is another interesting way to motivate the judgment about
b: As the example has been advanced, it is evident that the motion
of b has not changed during the closed interval from t = 1.5 to t = 2.
Nothing happened to b until after t = 2. Since b is traveling at two
meters per second at t = 1.5, and its motion is the same at t = 2, it
follows that b is moving at t = 2.
Some might object to this motivation, detecting a hint of circu-
larity. They will ask how we can judge that b’s state of motion at t =
2 is the same as it is at t = 1.5 without making a judgment about b’s
motion at t = 2, which is the point at issue in this example. But, in
fact, sensible as this objection sounds, we can make good judgments
of whether a property of an object has changed from one time until
another without antecedently judging much of anything about the
property at the later time. Suppose a philosopher interested in color
presented us with the following case: A purple billiard ball is placed
in what was an empty, very normally illuminated room at 12:02:00
P.M. The room is sealed shut with nothing inside but the ball and
whatever is providing the very normal illumination. No one can see
inside the room. The billiard ball is still in the room at 12:03:00
P.M. The philosopher interested in color asks: What color is the
 INSTANTANEOUS MOTION 53

billiard ball at 12:03:00 P.M.? Even though the specification of the

example does not explicitly say what color the ball is at 12:03:00
P.M., and even though no one in the hypothetical situation has any
direct way of checking its color right then, it is clear that the billiard
ball is purple at 12:03:00 P.M. As when we evaluate any hypo-
thetical example, we should assume that information not explicitly
specified is as normal and uninteresting as possible. The ball was
purple when it went in. There is no change of color brought on by
putting it in and leaving it in the room. Its color would not change.
Thus, we conclude that the ball is still purple at 12:03:00 P.M.
There is nothing circular about this judgment. If the philosopher
were considering some account of color that said about this case
that the billiard ball had no color (or some color other than purple),
this example would provide good reason to reject that account. Our
beliefs about what will change a property (be it a color or a motion)
shape our judgments about whether the property is still instantiated.5
It is easy enough to revise the standard definition of velocity to
avoid the unintuitive consequence about the obliterated object by
reworking the definition of a limit that it employs (cf., Tooley, p.
243). Instead of looking at the slopes of lines formed by the point
on the graph at t = t0 and other points on the graph to its left and
right, we could look at these slopes coming in from the left and
not the right. But, such a simple revision to the standard definition
subjects the Russellian view to other problems. It is a simple exer-
cise to construct a case of an object that is moving at t = 1 though
the object did not exist until t = 1. Think of a machine analogous
to Shoemaker’s (1984, p. 241) table producer creating an object that
comes into existence in otherwise empty space. If nothing happens
to it from t = 1 until t = 2, and at all times between t = 1 and t = 2 it
has a velocity of two meters per second, then the natural conclusion
to draw is that it was moving at t = 1. Here the temptation is to revise
the standard definition differently, looking at the slopes coming in
from the right and not the left!6
From the perspective of the Russellian view, what is needed
is some sort of analysis that lets the velocity of the object be
equal to the left-hand derivative if the left-hand derivative exists
and the right-hand one does not, the right-hand derivative if the
right-hand derivative exists and the left-hand one does not, but other-
54 JOHN W. CARROLL

wise lets the velocity be the value of the standard two-sided limit.
Unfortunately, this revised analysis will at least have some strange
consequences in cases where the limit is one thing coming in from
the left and something different coming in from the right. Consider
an object, c, whose position function is k(t) = 0 if t ≤ 1 and k(t) = 2t
− 2 if t > 1.

c starts out at rest. It remains at rest for a bit, from t = 0 until about t
= 1. After t = 1, c is moving. What is tricky about the function k is its
behavior at t = 1 (see Tooley, 1988, p. 245; Jackson and Pargetter,
1988, p. 141; Mortensen, 1985, p. 2). Like the original definition,
the revised definition leads to the conclusion that, on the Russellian
view, c is neither at rest nor in motion. The relevant limit does not
exist; it goes to two from one side, zero from the other. This is a
strange consequence for the revised analysis, because, if the object
had not existed before t = 1 but otherwise had this same position
function, then it would have been in motion at t = 1. But, since it did
exist before t = 1, it does not. Existing at earlier times undermines
being in motion at a later time. How strange. Strangeness aside,
additional problems for this revised Russellian view will be raised
in Sections 3 and 4.

2. IF THERE WERE NO CHANGE

In Section 1, I motivated judgments about b’s motion at t = 2 by

invoking considerations about what could have changed its motion.
 INSTANTANEOUS MOTION 55

In this section, I discuss object b again. I will support the judgment

that b is moving at t = 2 via an appeal to counterfactual consid-
erations about what b’s motion would have been like if certain
motion-altering events had not occurred.
Consider a situation involving a block of marble in an art studio
at 12:02:00 P.M. A sculptor comes in and chisels a chunk off of
the block at 12:03:00 P.M. Question: How much did the block of
marble weigh at 12:02:00? Answer: The block’s weight at 12:02:00
is what the block would have weighed at 12:03:00 if it had been
left alone. Though perhaps something only a philosopher would say,
this counterfactual answer is a true answer. If it is true that the block
would have weighed 300 kilograms at 12:03:00 P.M. had it been
left alone, then it is true that the block weighed 300 kilograms at
12:02:00 P.M. The weight of a marble block would not have changed
if it had been left alone from 12:02:00 to 12:03:00.
Question: What is b’s state of motion at t = 2? Answer: b’s state
of motion at t = 2 is what its state of motion would have been at t =
3 if it had been left alone, if, say, the object canceller had never been
fired. We judge this answer to be true because we judge motion in b’s
situation in the same way we judge weight in the block of marble’s
situation. Since it is pretty much a Newtonian world, and since there
are no other objects nearby, if b had not been obliterated, if it had
been left alone, then b would have continued on in a straight line
after t = 2, neither accelerating nor decelerating. Since b’s state of
motion at t = 3 would have been the same as it was at t = 2 and b
clearly would have been moving at t = 3, we should conclude that
b was moving at t = 2. Thus, counterfactual considerations support
the contention that the Russellian view does not correctly describe
the motion of b at t = 2.
It may appear that I have suggested a conceptual connection
between velocity and the counterfactual conditional. Some might
even wonder if there might be some counterfactual account in the
offing that might improve upon the standard definition of velocity.7
Indeed, what if we tried to turn my counterfactual heuristic into
a definition? What if we analyzed an object’s true instantaneous
velocity to be what its Russellian velocity would have been if it had
been left alone?
56 JOHN W. CARROLL

There are counterexamples that suggest that the counterfactual

heuristic cannot be turned into a new definition of velocity in the
way just suggested. For example, consider an inertial object d that
is in motion at t = 2 in an Aristotelian world, one where inertial
objects decelerate. What would its Russellian velocity be if it were
left alone? Well, prima facie, it is left alone and its Russellian
velocities at times after t = 2 are less than what its true velocity
was at t = 2. Similarly, we should ask whether an object’s change in
velocity could sometimes be a chancy feature of the object. At all
times before and including t = 2, an object could be traveling at two
meters per second. Then, purely as a matter of chance, it could have
a velocity of one meter per second at all times after t = 2. Since this
object is left alone and yet has a new velocity after t = 2, what this
object’s Russellian velocity would be like if it were left alone after t
= 2 would be no indication of what its true velocity was at t = 2.
These examples do not undermine the counterfactual heuristic
qua heuristic. In essence, the problem with defining velocity
counterfactually is parallel to the problem of specifying ideal condi-
tions in a counterfactual analysis of dispositions. It may well be
impossible to define an object’s fragility in terms of whether the
object would break if it were struck. But it does not follow from this
that there are no situations in which the fact that an object would
break if struck indicates that the object is fragile. Hence, though
I am skeptical about there being some successful counterfactual
definition of velocity, I stand behind the use of the counterfactual
heuristic to motivate certain judgments about motion.

3. SUPERNATURAL OR CHANCY POSITIONING

Bigelow and Pargetter describe a Malebranchean world where God

recreates an object and places it in the world at every instant
(1990, pp. 68-70). Unfortunately, it is not clear what this recreation
amounts to and whether there are the appropriate causal connections
to ensure the persistence of the object. Fortunately, the element of
recreation is not essential to their case.8 Dropping that element,
suppose instead that God is not much concerned about most of the
objects in the universe, but he is very concerned about one, object
e. He can see to it that, at any time, this object occupies any spatial
 INSTANTANEOUS MOTION 57

position in the universe. At noon, it might be on my desk; at noon

and one second, in Rome; and at any intermediate times, on Venus,
on Pluto or in the Andromeda Galaxy. Where the object is at each
instant is determined by one of God’s commands. Think of God as
making his commands by instantaneously entering a spatial location
onto a computer terminal exactly one second before his command
takes effect. Object e is where it is at every instant because that is
the position God typed in for it the second before. So God does not
do any recreating, merely some repositioning. Now, suppose that on
one occasion God decides to position e so that it at least appears to
be traveling at a rate of one meter per second; for one full second,
he instantaneously types in a continuum of spatial positions so that
the position function of e for a one second interval is l(t) = t. About
their own Malebranchean example, Bigelow and Pargetter say,
An object is created at one location at one time and at another location a moment
later. At neither time is it really moving – it has no genuine instantaneous velocity
. . . (p. 68)

Applied to my reworking of their example, this amounts to the

following: That e is not in motion and not at rest at t0 , where t0 is any
instant during that one-second interval. Of course, the Russellian
view (even with the revised definition of velocity explored at the
end of Section 1) will say that e is moving at t0 .
Are Bigelow and Pargetter correct that e is not moving and not
resting at t0 ? Our heuristic based on judgments of whether anything
happened that changed the object’s motion is not a useful heuristic
in this new case, at least not in the straightforward way it was useful
in Section 1; something rather dramatic is happening to the object at
every instant. But our counterfactual heuristic of Section 2 is more
enlightening. Let us consider what e’s state of motion would have
been later had it been left alone. What would have happened to e if
God had taken a break after positioning the object at time t0 minus
one second? On the one hand, if e would have traveled in a straight
line a distance of one meter in the next second, then it seems that the
Russellian view would correctly entail that it was moving at time t0 .
On the other hand, if e would have remained right where it was, it
seems that the Russellian view would incorrectly entail that it was
moving at t0 ; in that case, the object would be at rest. Thus, counter-
factual considerations of the sort raised in Section 2 suggest that
58 JOHN W. CARROLL

Bigelow and Pargetter have not quite correctly described the motion
of e at t0 . They think that the example is sufficiently specified to
determine the motion facts about e. As they see it, as the example
is specified, it is true that e is not in motion and not at rest. But,
in fact, the example is not sufficiently specified to determine the
motion facts. It is left open by the specification that the object could
either be at rest or in motion.
Tooley (1988, pp. 243–244) describes a similar example about
which it is at least more plausible to think that the object in question
lacks a true velocity. He asks us to suppose that, at every moment of
an object’s existence, its continued existence and its future spatial
locations (if it does continue to exist) are extremely chancy. No
matter what has transpired up to time t, this object is just as likely
to occupy many different spots in the universe in the moments
following t. According to Tooley:
In such a world, the locations occupied by an object might very well form a radi-
cally discontinuous “curve”. But, at the same time, it might occasionally be the
case that an object, during some interval, occupied positions all of which fell on
some continuous curve that was differentiable at every point. If this did happen,
Russell’s account would imply that the object possessed a velocity at every time
during that interval.
But is it true that an object, in such a probabilistic world, which just happened
to occupy positions on such a curve, would have a velocity? It seems to me that
one is very hesitant to attribute a velocity in such a case, and I would suggest that
the reluctance to do so derives from the feeling that the velocity of an object at a
time should be causally relevant to its positions at later times. (p. 244)

A bit later, Tooley recharacterizes his conclusion as “that a particle

in such a world would not have a velocity at any time, even if all
its positions happened to fall along a curve describable by some
continuous, differentiable function” (p. 244).
With Tooley’s example, matters are a little more tricky than they
are with my slight reworking of Bigelow and Pargetter’s example.
There is no omnipotent being doing anything to the object. Indeed,
there is no salient motion-changer that we might suppose away to
even begin to counterfactually get a grip on the object’s motion. In
Tooley’s case, the relocation of the object seems like it is governed
only by some fundamental probabilistic laws of nature. Are we to
counterfactually suppose that the laws of nature or at least one of
the laws of nature is suspended? If we do, would the behavior of the
 INSTANTANEOUS MOTION 59

object after the time in question be indicative of its motion? Perhaps,

Tooley has described a case of an object that truly is neither at rest
nor in motion. Perhaps not. The nature of the universe in his example
is such that I just do not know what to think.
While both my minor disagreement with Bigelow and Pargetter
and my discussion of Tooley’s case highlight the usefulness of
bringing in counterfactual considerations to a discussion of motion,
they do nothing to save the Russellian view. About Bigelow and
Pargetter’s case, we have seen that e may be in motion, but it could
also be at rest; the Russellian view (even with the revised definition
of velocity discussed at the end of Section 1) incorrectly insists that
e is in motion. With Tooley’s case, even if the object had a differ-
entiable position function, it looks like the object may neither be in
motion nor at rest. Yet, the Russellian view clearly says that it would
be one or the other. It is starting to look like a serious mistake to
assume that the spatiotemporal facts about an object will determine
its state of motion.

4. NONSUPERVENIENCE

Consider our object c whose position function is k(t) = 0 if t ≤ 1

and k(t) = 2t − 2 if t > 19 (see Section 1). According to the standard
definition of velocity and its revision at the end of Section 1, c has no
Russellian velocity. So, according to the Russellian view, c is neither
at rest nor in motion at t = 1. But, now, let us ask what c’s state of
motion really is. I contend that there are at least two possibilities
that are both consistent with everything that has been said about the
case so far. It might be that c is at rest at t = 1. Then again, it might
be that c is moving at t = 1.
World 1. Suppose the physical nature of the world is like this: When objects
interact, the effect of the interaction is present at the time of the interaction. So, if
a heated body comes in contact with a cooler body, and if the first moment they
are in contact is t0 , then there is a transfer of heat at t0 . c, we will suppose, is a
perfectly rigid sphere that is initially at rest. Another perfectly rigid sphere, the
same size as c, is rolling toward c at a rate of two meters per second. The rolling
sphere will contact c at t = 1.
World 2. Suppose that the physical nature of the world is a little bit different:
When objects interact, the effect of the interaction is present immediately
60 JOHN W. CARROLL

following the interaction though not at the instant of the interaction. So, for
example, and very roughly, if a heated body comes in contact with a cooler body,
and if the first moment they are in contact is t0 , there is no transfer of heat at t0 .
At all times following t0 , however, a transfer has transpired. With that in mind, let
us again suppose that c is a perfectly rigid sphere that is at rest and that the other
perfectly rigid sphere is rolling toward c at a rate of two meters per second. The
two spheres collide at t = 1.

Given the nature of interactions in World 1, the natural conclusion

to draw is that, at t = 1, c is in motion. Given how interactions
take place in World 2, the natural conclusion is that, at t = 1, c is
at rest. Thus, not only does the Russellian view fail, but whether an
object is moving does not even supervene on facts about the object’s
spatiotemporal locations.10
In formulating the nonsupervenience argument above, I have
asked you to suppose that, in World 1, the effect of an interaction
takes place at the time of the interaction. I have asked you to suppose
that, in World 2, the effects of interactions take place immediately
after the interaction though not at the instant of the interaction. This
together with a description of the shared particular circumstances in
the worlds seems to entail that c is at rest at t = 1 in World 2 and that c
is moving at t = 1 in World 1. But is this too quick? It might seem that
I have asked you to suppose exactly what friends of the Russellian
view would deny. I think, however, that a little more careful (though
much more laborious) presentation of the nonsupervenience argu-
ment would avoid any appearance of question begging. I do not
mean to stipulate about World 1 that the effects of all interactions
take place at the instant of the interaction. Being much more careful,
I would begin by describing several kinds of causal interactions.
In my actual presentation of the argument, starting with World 1, I
did describe how heat transfers take place. I could have also talked
about transfer of electric charge or any number of other types of
causal interactions. About all these interactions, I would ask you to
suppose that the effect of the interaction takes place at the instant of
the interaction. My hope would be that this would naturally lead you
to think that interactions involving simple motion transfers could be
like this too. If so, given my description of the particular facts in
World 1, you would naturally draw the conclusion that c is moving at
t = 1. Similar reasoning from appropriately different descriptions of
these other interactions in World 2 should lead you to conclude that
 INSTANTANEOUS MOTION 61

it is possible that the described particular circumstances that occur

prior to t = 1 in World 2 are compossible with c being at rest at t
= 1. There is nothing tied to the Russellian view that does anything
to suggest that heat transfer, transfers of charge, or what have you
could not be as I would have described them in a more extended
description of World 1 and World 2. And its from these stipulations
that one is to infer the possibility of worlds like World 1 and World
2. No question has been begged against the Russellian view.
Is there more that can be said in support of the judgments (i)
that, in World 1, at t = 1, c is in motion and (ii) that, in World 2,
at t = 1, c is at rest? I believe there is. Reasoning of the sort used
in Section 1 strongly supports these conclusions: In World 1, after
t = 1, c is clearly traveling at a rate of two meters per second. It
is also clear that there is nothing that happens during the interval
following t = 1 that would alter c’s motion. Thus, in World 1, c must
have the same state of motion at t = 1 that it does in the interval
following t = 1. Therefore, in World 1, c is moving at t = 1. In World
2, nothing has any effect on c during the closed interval from some
time just before t = 1 to t = 1. The effect does not obtain until after
t = 1. So, we naturally think there is no change in its motion from
intervals preceding t = 1 to t = 1. Since it is clear that c is at rest in
the intervals preceding t = 1, we conclude about World 2 that c is at
rest at t = 1. Furthermore, counterfactual reasoning of the sort used
in Section 2 bolsters my conclusion about World 2: In World 2, c is
affected after t = 1, but it is plausible to think that, if it were not, then
it would have covered zero meters in the next second. If that rolling
ball had not struck c at time t = 1, if there had not been any effect
on c after t = 1, then c would have remained right where it was. Its
velocity from t = 1 to t = 1 + δ would not have changed and at t = 1
+ δ it would have been at rest. Therefore, in World 2, at t = 1, c is at
rest.
The World 1 and World 2 examples show that an object’s spatial
and temporal positions do not always determine whether it is
moving. I have described two possible situations. In one, an object
has a certain set of spatial and temporal locations and is in motion.
In the other, this same object occupies the exact same set of spatial
and temporal locations and yet the object is at rest. No account
of instantaneous motion stated just in terms of an object’s spatial
62 JOHN W. CARROLL

and temporal locations could be true about both of these possible

situations.

5. THE UNREPROACHABILITY OF THE RUSSELLIAN VIEW

The Tooley and Bigelow-and-Pargetter counterexamples have not

caused much of a stir among philosophers. Nor do I expect my
revamping of their examples to change this situation very drastically.
In this section, I will reveal several curious features of the present
issue that conspire against vanquishing the Russellian view, despite
the revealed shortcomings.
One reason for the Russellian view’s permanence is that it
suggests a sensible epistemological method for judging whether an
object is in motion or at rest. Suppose there was a property that could
be possessed at time t0 without being had at any earlier times or at
any later times. Given that you could not directly observe that an
object had this property at t0 , what would be the best way of trying
to find out whether it did? Well, you would observe the object during
intervals that include t0 . It is natural to expect that observations
about smaller and smaller intervals around t0 would be better and
better evidence of the object’s state at t0 .
Another reason the Russellian view survives is that, for all that I
have argued, the connection it describes between motion and the
first derivative of the position function may not be far from the
truth. The central thesis of my paper is not a challenge to the claim
that an object is moving at a time if and only if the value of the
derivative of the object’s position function at that time is nonzero.
Instead, my central thesis is a challenge to the claim that motion
can be defined in this way. To have given a successful definition
of instantaneous motion in terms of the derivative of the position
function, it must be true that it is necessarily true that an object is
moving at a time if and only if the value of the derivative of the
object’s position function at that time is nonzero. Science fiction
examples, so long as they are genuinely possible, can show this
claim to be false without establishing anything about how motion
takes place in the actual world. The introduction of perfectly rigid
spheres, omnipotent relocators, and so forth ensure that most of my
examples are purely hypothetical.
 INSTANTANEOUS MOTION 63

Despite appearances, the obliterated-object case and the created-

object case discussed in Section 1 are somewhat realistic. Of
course, I realize that there are no Shoemakerian object producers
or cancellers, but all that is crucial about these examples is that
some moving object have a first or a last moment of existence. These
cases are trouble for the Russellian view when accompanied by the
standard definition of velocity. But, and this is the important point,
they are no threat to the Russellian view when combined with the
revised definition of velocity from the end of Section 1. My only
potentially realistic counterexample with the revised definition in
play was also given at the end of Section 1. It involved object c,
which has a continuous position function with a point where the
left- and right-hand derivatives both exist and are not identical. As
far as I know (and, I think, as far as anyone knows – see below),
this may also take us into the realm of science fiction, leaving it
open that the Russellian view combined with the revised definition
of velocity is actually true. Thus, as far as anything I am prepared
to argue is concerned, this variation of the Russellian view may be
a true, perhaps even a nomically true, description of the relationship
between an object’s motion at an instant and its position function.
That may be enough for physics to rely on the Russellian view in its
explanations of the motions of objects.
Let me offer one last and more speculative reason for the
unreproachability of the Russellian view. As we all know, there
are some unknowable propositions. Work on vagueness tells us
that there are borderline propositions that we couldn’t know. There
are also Moorean propositions like that it is raining but I don’t
believe it. The class of unknowable propositions that are of partic-
ular interest to me are those whose knowledge requires us to notice
differences well beyond the capabilities of any human to notice
(cf., Sorensen, 1988, pp. 68–69). These include propositions like
that Mount Rainier is exactly 4392 meters high (even assuming it
is exactly 4392 meters high) or that the Rosetta Stone was found
exactly 200 years ago (even assuming it was found exactly 200 years
ago). The problem is that it is hard to see how any being with finite
capacities of discrimination could know that the Rosetta Stone was
found 200 years ago and not even slightly less (or slightly more)
than 200 years ago. How could anyone know that Mt. Rainier is
64 JOHN W. CARROLL

4392 meters high and not even a slight bit higher, 4392 meters high
and not even a slight bit less high?11
These cases are interesting because the counterexamples to the
Russellian view all involve some object’s having certain properties
at exactly one time but not at times arbitrarily close to that time.
So, even if there are objects like this in the actual world, we cannot
know any such information about them. For example, we could not
know that object b existed at t = 2 and did not exist at all times after t
= 2. There is no way we could distinguish that situation from a situ-
ation in which b exists until t = 2 + δ seconds for some positive and
sufficiently small δ. Or, let us think about object c. We are told that
it occupies a different position at each time after t = 1, but occupies
the same position at the times leading up to and including t = 1. It
is these propositions that help comprise the hypothetical example.
This conjunction of propositions is an unknowable proposition.
Thus, in addition to conforming to a sensible epistemology and to
admitting of a revision that may be true (without being necessarily
true), the Russellian view has a shield generated by our epistem-
ological limitations. The counterexamples would be in blindspots
if they were actual. We can know that the Russellian view fails
as a definition, because we can specify hypothetical situations that
encourage judgments about motion at odds with that account. We
make these judgments based on propositions that are part of the
specification of the example, and that is enough to judge that the
Russellian view is not a necessary truth. We cannot, however,
actually know these specified propositions to be true (true in the
actual world) because that would require powers of discrimination
well beyond us. So, at least as far as the Tooley and Bigelow-
and-Pargetter examples are concerned, the Russellian view can be
employed as a contingent truth without fear that someone will
actually stumble upon a situation that could be known to threaten
it.

ACKNOWLEDGEMENTS

A version of this paper was read at the colloquium honoring David

Sipfle on the occasion of his retirement from Carleton College. The
High Priest, as he is known by his loyal following of Zenophiles,
 INSTANTANEOUS MOTION 65

introduced me to the topic of motion and thereby hooked me on

philosophy. An excerpt of a more recent version of this paper was
read at the year 2000 joint meetings of the North Carolina and
South Carolina Philosophical Societies. Thanks to Randy Carter,
Hal Levin, David MacCallum, Dan Schroeder, Roy Sorensen, and
especially Michael Casper, John Roberts and Doug Jesseph for
providing many helpful comments and lots of fun discussion.

NOTES
1 Conversation being what it is, there are some implications of my saying what
I did, implicatures to the effect that I will remain in my office long enough to
accomplish whatever needs to be done. These implicatures, however, are not part
of the content of my statement.
2 It is a presupposition of the Russellian view that space and time are continuous

in the mathematical sense; I will not question that presupposition here. In fact, all
the examples to be discussed involve objects that exist in an absolute Newtonian
spacetime (except where otherwise indicated). All my discussion of motion will
be restricted to linear motion.
3 Denis Robinson, 1989; Dean Zimmerman, 1998, 1999; Katherine Hawley,

1999; David Lewis, 1999; Craig Callender, 2001.

4 I have constructed this example based on Tooley’s discussion (1988, p. 243)

(also see Bigelow and Pargetter, 1990, p. 67).

5 There is a history of employing this philosophical method. Consider a horribly

simplistic idealism holding that existence requires perception by some particular

perceiver. It is natural to object to such a view that when this perceiver leaves a
room “all the furniture in the room is reduced to nothing” (Berkeley, Principles,
sec. 45, p. 42). This quote is from a passage where Berkeley is considering
objections to his idealism. He makes it clear that his view is not so simplistic.
“Wherever bodies are said to have no existence without the mind, I would not
be understood to mean this or that particular mind, but all minds whatsoever”
(Principles, sec. 48, p. 44). But what is the objection exactly? We are asked to
suppose that the perceiver is in a room looking at the furniture. Now suppose he
leaves the room. Does the furniture still exist? Of course, the natural and correct
response to make to the example is that the furniture does continue to exist. But
doesn’t this beg the question against the simplistic idealism? Isn’t the existence of
the furniture when it is not perceived what is at issue? The history of philosophy
has rightly judged that the question has not been begged. We judge that nothing
has happened to the furniture and are very confident that when nothing happens to
furniture it continues to exist. Respecting these judgments, we reject the simplistic
idealism.
6 Rigorous definitions of the two-sided limit used in the standard definition and

the one-sided limits discussed here can be found in (Anton, 1992, pp. 132, A28).
66 JOHN W. CARROLL

7 The 18th century mathematician Colin Maclaurin did define velocity subjunct-
ively by defining uniform motion in terms of covering equal distance in equal
amounts of time and then defining instantaneous velocity in terms of the distance
the object would travel if it moved uniformly from the time in question (see
Jesseph, 1993, pp. 281–282). But, Maclaurin’s approach gives an incorrect judg-
ment about the supernatural positioning case to be discussed in Section 3.
According to MacClaurin’s definition of uniform motion, the object is moving
uniformly from the time in question. So his approach gives the same incorrect
result as the Russellian view; it says that e is in motion. Maclaurin’s approach
also gives a mistaken judgment about the World 2 example to be presented in
Section 4, judging that c has a velocity of two meters per second at t = 1 in World
2.
8 Bigelow and Pargetter initially talk about the velocity of an image on a movie

screen and it appears that the Malebranchean world is supposed to be analogous

(p. 68). After raising sensible concerns about whether an image is a persisting
object at all, they make it clear that the recreation is suppose to be consistent with
the original object persisting (p. 70). By removing the element of recreation, we
can lessen these concerns further.
9 Tooley (1988, pp. 246–247) also uses this example to argue for nonsuperveni-

ence.
10 As described, World 1 and World 2 depend on there being some sort of

instantaneous causation. While I think such causation is possible, this is an

incidental feature of the examples. Both examples work equally well if the trans-
ference of the velocity takes some time. In World 2, it could be that if bodies
interact at t0 then there is no change in the affected body until after one milli-
second has passed. So, at t0 + 1 millisecond, the second body is still unaffected
but at all times after t0 + 1 millisecond it is affected. Then, if c is struck at exactly
one millisecond before t = 1, then it is still at rest at t = 1. In World 1, it could be
that if bodies interact at t0 then the effect obtains exactly one millisecond later.
Then, if c is struck at exactly one millisecond before time t = 1, then it is plausible
to think that c is moving at t = 1.
11 No one knows that Mt. Rainier is 4392 meters high? Have we wandered into

skepticism? Well, we have to be careful. I am sympathetic to the claims of Peter

Unger (1986), Stewart Cohen (1988), Keith DeRose (1995), and David Lewis
(1996) to the effect that the verb ‘to know’ exhibits a high degree of context-
sensitivity. It is plausible to think that there are contexts where it is true to say,
‘I know that Mt. Rainier is 4392 meters high’. The context generated by the
paragraph in the text is not one of them.

REFERENCES

Anton, H. (1992): Calculus, New York: Wiley and Sons.

Aristotle (1941): The Basic Works of Aristotle, New York: Random House.
 INSTANTANEOUS MOTION 67

Berkeley (1965): Principles, Dialogues, and Philosophical Correspondence.

Indianapolis: Bobs-Merrill.
Bigelow, J. and Pargetter, R. (1990): Science and Necessity. Cambridge:
Cambridge University Press.
Callender, C. (2001): ‘Humean Supervenience and Rotating Homogeneous
Matter’, Mind 110, 25–44.
Cohen, S. (1988): ‘How to be a Fallibilist’, Philosophical Perspectives 2, 91–123.
DeRose, K. (1995): ‘Solving The Skeptical Problem’, Philosophical Review 104,
1–52.
Hawley, K. (1999): ‘Persistence and Non-Supervenient Relations’, Mind 108, 53–
67.
Jesseph, D. (1993): Berkeley’s Philosophy of Mathematics, Chicago: University
of Chicago Press.
Jackson, F. and Pargetter, R. (1988): ‘A Question about Rest and Motion’,
Philosophical Studies 53, 141–146.
Lewis, D. (1996): ‘Elusive knowledge’, Australasian Journal of Philosophy 74,
549–567.
Lewis, D. (1999): ‘Zimmerman and the Spinning Sphere’, Australasian Journal
of Philosophy 77, 209–212.
Mortensen, C. (1985): ‘The Limits of Change’, Australasian Journal of Philos-
ophy 63, 1–10.
Robinson, D. (1989): ‘Matter, Motion, and Humean Supervenience’, Australasian
Journal of Philosophy 67, 394–409.
Russell, B. (1903): The Principles of Mathematics, Cambridge: Cambridge
University Press.
Shoemaker, S. (1984): Identity, Cause and Mind, Cambridge: Cambridge Univer-
sity Press.
Sorensen, R. (1988): ‘Vagueness, Measurement, and Blurriness’, Synthese 75, 45–
82.
Tooley, M. (1988): ‘In Defense of the Existence of States of Motion’, Philosoph-
ical Topics 16, 225–254.
Unger, P. (1986): ‘The Cone Model of Knowledge’, Philosophical Topics 14,
125–178.
Zimmerman, D. (1998): ‘Temporal Parts and Supervenient Causation: The Incom-
patibility of Two Humean Doctrines’, Australasion Journal of Philosophy 76,
265–288.
Zimmerman, D. (1999): ‘One Really Big Liquid Sphere: Reply to Lewis’,
Australasian Journal of Philosophy 77, 213–215.

Department of Philosophy and Religion

North Carolina State University
Raleigh, NC 27695-8103, USA
E-mail: carroll@unity.ncsu.edu