Sorabji-Kretzmann, Aristotle On The Instant of Change

Aristotle on the Instant of Change
Author(s): Richard Sorabji and Norman Kretzmann

Reviewed work(s):
Source: Proceedings of the Aristotelian Society, Supplementary Volumes, Vol. 50 (1976), pp.
69-89+91-114
Published by: Wiley-Blackwell on behalf of The Aristotelian Society
Stable URL: http://www.jstor.org/stable/4106825 .
Accessed: 05/10/2012 01:42
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Wiley-Blackwell and The Aristotelian Society are collaborating with JSTOR to digitize, preserve and extend
access to Proceedings of the Aristotelian Society, Supplementary Volumes.
http://www.jstor.org
ARISTOTLE ON THE
INSTANT OF CHANGE
Richard Sorabji and Norman Kretzmann
I--RichardSorabji
The problem
"The train leaves at noon", says the announcer. But can it?
If so, when is the last instant of rest, and when the first in-
stant of motion? If these are the same instant, or if the first
instant of motion precedes the last instant of rest, the train
seems to be both in motion and at rest at the same time, and is
not this a contradiction? On the other hand, if the last instant
of rest precedes the first instant of motion, the train seems to
be in neither state, during the intervening period, and how
can this be? Finally, to say there is a last instant of rest, but
not a first instant of motion, or vice-versa, appears arbitrary.
What are we to do?
This kind of problem has had a long history. It is to be
found already in Plato's Parmenides (156C-157A), and it had
a great revival in mediaeval times.' To this day, it remains
relevant to the definition of motion at an instant, as we shall
see. I want to suggest a solution to it, which turns on treating
motion differently from rest. I want then to argue, that though
Aristotle has been castigated for attempting to solve the prob-
lem by means of a mistaken denial of motion or rest at an
instant, he was also attracted by a second solution very close
to the one I shall argue for. His treatment of dynamics can to
this extent be reassessed.
Does it apply to the real world?
First we need to consider whether the problem could apply to
the real world. It may be doubted whether it could, for the
statement of the problem involved a number of assumptions.
First, I assumed (what Aristotle argues in the Physics) that
time is continuous. This has many implications. It means that
time will be infinitely divisible, and there will be no such
70 I-RICHARD SORABJI
thing as a time-atom, that is, an indivisible period with an

indivisible duration. An instant will be not a time-atom, nor
any kind of period, but rather the boundary of a period, itself
having no duration. Instants, unlike time-atoms, cannot be
next to each other (cf. Phys VI 1, 232a 6-11). Rather, between
any two instants, there will be another, indeed, an infinity of
others. This is what is involved in time being continuous, and
our problem will apply to the real world only if time is so. If
there were time-atoms, so that time was not continuous, the
train would be in its old postion at one time-atom, and in a
new position at the next. The earlier time-atom would be the
last time-atom in the period of the train's resting; the later
time-atom would be the first time-atom in the period of the
train's moving, and our problem would not arise. The prob-
lem does arise, however, if Adolf Griinbaum is right that
neither quantum theory, nor anything else in modern physics,
has given us reason to deny that time is continuous (Modern
Science and Zeno's Paradoxes, Middletown, Connecticut
1967).
Another reason why someone might question whether our
problem applies to the real world is that a train consists of a
mass of moving atoms, and so does the railway track. Can the
train have any first instant of motion, or last of rest, if its
atoms are moving all the time, and how would these instants
be defined? Yet another doubt concerns the fact that a train
is not perfectly rigid. When some parts of the train, or of the
engine, have started to move, other parts will be lagging be-
hind, so that there is not a single first instant of motion or last
of rest for the train as a whole. Both these doubts can be met
by raising our problem not about the train as a whole but
about some point within the train, such as the centre of mass,
and its first instant of motion and last of rest, in relation to
some point on the railway track.2In talking of points, rather
than of trains, we will be moving beyond the range of observ-
able entities.
So far as I can see, then, our problem does apply to the real
world, in as much as it applies to unobservable points on a real
train. But two further things need to be said. First, the prob-
lem would still be of interest, even if it applied only to a world
different from ours. Second, we have so far considered only
ARISTOTLE ON THE INSTANT OF CHANGE 71
one version of the problem, and if this version were inapplic-

able to the real world, it might still be the case that other
versions were applicable. Thus far I have considered only the
transition between rest and motion. But our problem can be
raised, and was raised by Aristotle, in connexion with other
kinds of transition. He discusses the transition from being
one colour to being another colour, from being non-existent
to being existent, and from being invisible to being visible.
(The last case is considered in De Sensu 7, 449a 21-31, and it
has interesting connexions with the ancient paradox of the
heap or bald man; I regret that space prevents me from dis-
cussing it.) In each case, the question can be asked: when is
the last instant of the old state, and when the first instant of
the new? Or moving from time to space, we may be able to ask
where is the last point of the one state, and the first point of
the other? In some of these new forms, the question may well
apply to our world.
Proposed solution for casesof continuous change

With the problem now stated and generalized to apply to all
kinds of transition, we can start making some suggestions
about how to handle it. But first we should be clear how much
we need to ask of a solution. The original question was about
when the last instant of the old state occurs, and when the first
instant of the new. One of the difficulties about answering was
that if we said that one of these instants existed, but not the
other, we seemed to be being arbitrary. It would be a sufficient
solution, if we could show that it would not be arbitrary to
prefer one instant to the other. For this purpose, we need only
show that there is a reason for preferring one to the other; we
need not show that it is mandatory to do so. On this basis, I
would suggest that there is a solution available for those cases
where the earlier state, or the later state, or both, consists in a
continuous change. We can illustrate by considering a transi-
tion from rest to motion and back again to rest, provided that
the motion is construed as continuous, not jerky. By this I
mean that the motion involves passing through an infinity of
points, between any two of which there are other points, which
are also passed through. Ordinary usage is not precise, but
leaves it indeterminate whether we should regard the instant
of transition between rest and continuous motion as an instant

of motion or not. We must therefore make a recommendation
about how to regard it, if we want to solve our problem. Fortu-
nately, there are several considerations which would justify the
decision to call it an instant of rest, and I shall mention three.
First, there is an asymmetry between the series of positions
away from the position of rest and the position of rest itself.
There can be no first position away from the starting point, or
last position away from the finishing point in a continuous
motion, or in any other continuous change. Hence there can
be no first instant of being away from the starting point or
last instant of being away from the finishing point. No such
considerations apply to being at the position of rest. This al-
ready supplies us with a solution to our paradox, in some of its
applications. For if someone were to ask. "when is the last
instant of being at the position of rest, and when the first of
being away from it?", we could safely reply that the latter
instant does not exist. But we can go further. The asymmetry
between the position of rest and the positions away from it can
provide us with the excuse we want for treating rest
differently from motion. It would be perfectly reasonable to
mark the asymmetry by saying that just as there is no first or
last instant of being away from the position of rest, so equally
there is no first or last instant of motion. It would be reason-
able, but not mandatory. Reasonableness is all we need in
order to escape the charge of arbitrariness.
This decision is reinforced by two further considerations.
First, let us suppose that not only change of place is continu-
ous, but also change of velocity. In other words, in passing
from one velocity to another, an object passes through all the
infinitely many intervening velocities. Once again, quantum
theory has nothing to say against this assumption. If it is made,
then there cannot be a first or last instant of having a velocity
greater than zero, for there is no first or last velocity above
zero. There is, however, no corresponding objection to there
being a first or last instant of having velocity zero. Now it
seems more natural, though again it is not mandatory, to con-
nect zero velocity with rest, and velocities above zero with
motion. If we do, we get the result that there is no first or last
instant of motion, but that there may be first or last instants of
rest.
Further support for this decision is gained, if we consider a

special case of coming to a halt. We can imagine the centre of
mass of a ball travelling vertically upwards and slowing down
until it reverses direction at an instant, without pausing for
any period of time at the apex of its journey. We will now have
an extra incentive for denying that the centre of mass is in
motion at the instant at which it is at the end of its upward
journey. For not only will its velocity be zero at that instant,
but we could not say that its motion had one direction rather
than the opposite direction at that instant. It will be easier,
then, to describe the instant of reversing direction as one of
rest. And if we do, there can be no last or first instant of
motion on either side.3
In arguing for the non-arbitrarinessof the decision to deny
first and last instants of continuous motion, we are not saying
that the period of motion has no boundary. It will have an
instant bounding it on either side, and our only question has
been whether that instant should be regarded as one at which
there is motion, or rest. We need not worry, incidentally, if a
reason is found for saying that rest and motion are not contra-
dictories, and that the instant of transition between rest and
motion is neither one of motion nor of rest. We have found no
such reason, but if one were found, the resulting view would
not be so very different from ours, since at least it would treat
the instant of transition as one of non-motion.
I do not deny that there are considerations which point to
the opposite decision, that the instant of transition is after all
one of motion, and it will be as well to bring some of them into
the open. First, at the instant of reversing direction, a ball's
centre of mass is (as in all cases of coming to a halt) at a
different position from that occupied at preceding instants,
and (differently from ordinary cases of coming to a halt) it is
also at a different position from that occupied at succeeding
instants. This difference of position admittedly favours our
regarding the instant of reversing direction as one of motion.
But I think the consideration is outweighed by the absence
of a particular direction and of a positive velocity.
A further doubt which might be raised is whether change
of acceleration always behaves in the same way as change of
velocity. We have supposed that velocity always changes con-
tinuously. But does acceleration, the rate of change of velocity,
always in its turn change continuously, or can it jump dis-

continuously from zero to, say, one foot per second per second?
A similar question could be raised about yet higher deriva-
tives, such as the rate of change of acceleration. Nonetheless, I
do not think our solution is seriously threatened. On the one
hand, if acceleration does change continuously, there will be
no first instant of acceleration above zero, and so we will have
yet further incentive to deny a first instant of motion. On the
other hand, if acceleration were sometimes to jump discon-
tinuously from zero to something higher, we should admittedly
have no obvious general reason for choosing between talk of
a last instant when acceleration is zero and a first instant when
acceleration is above zero. But even if on some of these occa-
sions we were to talk of a first instant of acceleration above
zero, the considerations we rehearsed earlier would still be
strong enough to make us hesitate before calling that first
instant of positive acceleration an instant of motion.
In spite of this defence, our proposal must be understood
in a flexible spirit. We should recognize that for everyday
purposes, and for many scientific purposes, it simply does not
matter which way one talks, One's choice can legitimately be
based on the most transient of reasons, or on no reason at all,
while the reasons we have given can without penalty be
ignored. The point of our reasons is simply that they are
available to rebut the charge of arbitrariness in case of need.
If a discontinuous jump from zero acceleration were to occur,
and if in the context our whole interest were in the accelera-
tion to the exclusion of position and velocity, and in the posi-
tive acceleration, rather than in the zero acceleration, then
our reasons for denying a first instant of motion might be over-
ridden in that particular context; and this would not matter.
The point is that we would still not be forced to be arbitrary.
The solution suggested does not in any way preclude
physicists' talk of initial velocity. For the initial velocity of a
projectile is not a first velocity in its entire motion, but merely
the first velocity which it is convenient to consider for the
purposes of a given calculation. We should also recognize that
the last instant of rest in relation to one point may, of course,
be an instant of motion in relation to a different point.
Throughout we must be understood as talking of rest or
motion in relation to a given point.
The solution will apply not only to continuous motion, but

to all changes which are continuous in the same sort of way. It
will apply to changes of size, of temperature, or of colour, if
these involve a continuous progress through a series of points
between any two of which others are traversed. It will not
apply to discontinuous changes, or to other processes or states
which lack this kind of continuity. If, for example, we con-
sidered the transition from not singing to singing, we should
not have the same reasons to deny a first instant of singing.
Clearly, arguments based on direction or velocity would be
inapplicable. It might be thought that something like our first
argument would still be applicable, for the singing lasts
through a continuous series of instants, and there can be no
first instant after the instant of transition. But our first argu-
ment cannot in fact be applied, for it depended on the
asymmetry between the single position occupied during a
period of rest and the continuous series of positions occupied
during a period of motion. In the case of singing, there is no
asymmetry, for there is a continuous series of instants
traversed during the period of non-singing, just as much as
during the period of singing.
One might expect the proposed solution to appeal to
Aristotle, for in his attack on the atomists he is at great pains
to insist that motion, time and space are all alike continuous.
He argues hard against the atomists that what has moved must
previously have been moving (Phys VI 1, 232a6-18; VI 6
237a17-b22; VI 10o 240b31-241a6); it cannot simply have
jerked into its new position (232a6-11; 24ob31-241a6). It was
his successorsDiodorus Cronus and Epicurus (or his followers)
who were willing to accept jerky motion. (For Diodorus, see
Sextus Empiricus, adv. math. 10.48; 85-6; 91-2; 97-o102; 143;
cf 12o. For Epicurus or his followers, Themistius, Phys 184.9;
and Simplicius, Phys 934-24.) Nonetheless, we shall later see
that Aristotle's solution is somewhat more complicated.
Treatment of other cases

Our solution leaves a very large range of cases unsolved. For
often neither the earlier state nor the later state is a process
of continuous change, as we have remarked. The transition
may be from one colour to a different colour, from non-
76 I --RICHARD SORABJI
existence to existence, or from invisibility to visibility. In

these cases, what considerationsare there, to help us to a
decision?If we are watchinga recedingaeroplane,or looking
for an approachingone, we cannot normallytell at the time
whatwill proveto be the last instantof visibility as it recedes,
or the last instant of invisibility as it approaches.If we want
to registerthis instantas it arrives,we shall normallyhave to
wait until the new state is upon us, before we can do so, and it
may then reasonably be held that we are not registering the
end of the old state, but, at best, the beginning of the new.
This means that, in many contexts, we have a good reason
for not talking of the last instant of the old state, but (if it
has one) of the first instant of the new. This solution seems
to have appealed to Peter of Spain, for certain kinds of case.4
But it needs to be noticed that our interest is not always in
registering the instant as it arrives. We may instead want to
discuss the instant of transition prospectively or retrospec-
tively. So the present consideration does not provide a solution
for all cases, or even for all the cases Peter of Spain applies
it to.
Aristotle himself may have another consideration relevant
to the particular example of visibility. For he classifies seeing
as an energeia, and on one interpretation, an energeia has no
first instant. This is how J. L. Ackrill interprets Aristotle's
idea (e.g., Sens. 446b2) that "he is seeing" entails "he has
seen". Ackrill treats the perfect tense "he has seen", like "he
has been seeing", as implying an earlier period of seeing
("Aristotle's distinction between Energeia and Kinesis", in
New Essays on Plato and Aristotle, ed. R. Bambrough,
London 1965, esp. pp. 126-7). This interpretation has been
disputed, but if it is correct, it implies that there will not be a
first instant of seeing, and therefore not a first instant of seeing
the approaching object in the problematic example of De
Sensu 449a21-3 1, which was briefly mentioned above, p. 71.
The various considerations we have mentioned do not begin
to cover all the cases there are. There may well be unique con-
siderations attaching to particular occasions of discussion. And
we must add that there may well be cases in which there are
no adequate considerations to guide us. In these last cases, we
shall not be able to answer the question, "what is the last
instant (or point) of the one state and the first of the other?"
The most we shall then be able to do is this. If our questioner
happens to assume (without reason) that one of the instants (or
points) exists, we shall always be able to tell him that in that
case the other does not exist.
A rival solution
An entirely different way of trying to cope with our problem
has been advocated by Brian Medlin in his paper "The Origin
of Motion" (Mind 1963). Medlin says, in effect, that a thing
can be both in motion and at rest at an instant, and equally
neither in motion nor at rest at that instant. The first may
sound as if it violates the law of contradiction, the second as if
it violates the law of excluded middle. But Medlin avoids this,
by simply defining motion at an instant, and rest at an instant,
in such a way that they are neither contradictories nor con-
traries of each other.' Given his definitions, all four statements
can be true together, namely, that a thing is in motion at an
instant, not at rest at that instant, and that it is at rest at that
instant, not in motion at it.
My objection to this is not so much that it runs the risk of
causing confusion, but that it is not sufficient to solve the
problem that interests us. Medlin is free to define motion at an
instant and rest at an instant in such a way that they are not
contradictories or contrariesof each other. But he cannot, and
does not, deny that there is a contradictory of the claim that
something is in motion at an instant. He himself suggests a
way in which we might formulate the contradictory.We could
talk of its being the case that something is in motion at a
certain instant, and of its not being the case that something is
in motion at that instant. Once we have found a formula for
picking out the contradictory,we can pose our original prob-
lem all over again in terms of the new formula. We shall
simply ask what is the last instant when it is not the case that
our object is in motion, and what the first instant when it is
the case that it is in motion. To this question Medlin himself
would agree that we cannot say it is the same instant. When
the problem is posed this way, we see that we shall have to
fall back on a different solution from Medlin's, such as the
one we have advocated, according to which there can be a
last instant when it is not the case that our object is in motion,
but not a first instant when it is the case that it is in motion.
Aristotle's treatment. Preliminaries (i): The four main kinds

of change and the thesis that they all involve a gradual
transition.
I shall now turn to Aristotle's solutions of the problem. He is
well aware that different kinds of case need different solutions,
but, not surprisingly, he looks for solutions of some generality,
and does not acknowledge that the matter might ever be
decided by the unique interests of a particular context. We
shall suggest that in the cases we are going to discuss, Aristotle
is attracted by two solutions. Sometimes, like us, he denies that
certain continuous changes can have a first or last instant,
without however, making it very clear that it is the continuity
which precludes this. At other times, he goes further, and
denies that there can be any change or stability at an instant.
Before expounding his two solutions, we shall have to make
some preliminary points clear. A first thing to notice is that
Aristotle recognizes only four kinds of change as being changes
in the full sense of the word. There is change of quality (as
when something changes colour), change of place (in other
words, motion), change of size (in growth and diminution),
and finally the creation or destruction of substances (Phys III
1, 200oob32-2o0a16; cf V 2; Metaph XI 12).
A second point is that in all four kinds of change, he thinks
there is a gradual process of transition (Phys VI 6, 237a17-b3;
b9-21). Qualitative change, such as change of colour, is said to
take time. Change to a new place or size involves passing
through intervening points. The creation of something like a
house takes time, and occurs part by part, the foundation
before the whole.
Aristotle does not by any means think that changes other
than the four genuine ones must all involve a gradual process
of transition. Indeed, he sees that an infinite regress would be
involved, if the gradual process by which something came
into being had itself to come into being by a gradual process
(Phys V 2, 225b33-226a6). In the very passage where he ex-
plains that a house comes into being only part by part, he
points out that this cannot be true of things that have no parts
(VI 6, 237b 11, b15), points and instants, for example.
Preliminaries (ii): How can change of colour be gradual?

It may be wondered how change of colour can be gradual,
given the view stated in the De Sensu (6, 445b21-9g;
446a16-2o), and alluded to in the Physics (VI 5, 236b5-6;
b 7-18), that there is only a finite number of discriminable
shades. In that case, a change to a new colour would be gradual
when it involved passing through a number of intervening
shades. But within this process, how could the change from
one shade to the next be treated as gradual?
Aristotle seems to have two incompatible answers. One
answer is implied in Physics V 6 (23ob32-231a1), VI 4
(234bio-2o), VI 9 (24oa19-29), VI 1o (24ob21-31), when he
says that certain changes occur part by part.6While a surface is
changing from white to the next shade, grey, he says, part of
the surface must still be white, and part already grey. The
greyness spreads gradually over the surface. This claim, that
certain changes occur part by part, is used in combating the
view that a partless atom could undergo change or motion. In
its turn, the claim has as its ground that, while a thing is
actually changing or moving, it cannot yet be in its terminal
state, nor can it still be in its initial state. It must therefore
be partly in one state, partly in the other, and so must have
parts. In the case of motion, it must move part by part into
the adjacent area. Diodorus Cronus and certain Epicureans
were later to get round this objection to the motion of partless
atoms, by denying that an atom actually is moving at any time;
rather, at any given time it has moved with a jerk (references
above at the end of section on Proposed Solution, p. 75)-
But Aristotle seems not always to keep in mind the view that
these changes occur part by part. For in the De Sensu (6,
447a1-3) he actually denies that qualitative change has to
occur part by part, and illustrates how it can happen with the
case of a whole pond freezing over at once. In one of the
Physics passageswhere he says that qualitative change (en tois
enantiois 237b1) is gradual, he speaks as if he must prove this
on the basis of time taken, but cannot prove it on the basis of
space covered (237a19-28; a29; b2; b21). Why not, if he re-
members his view that a surface changes colour part by part?
He seems to be forgetting that view, and he probably forgets it
again in VIII 8, where he declares that while something is
becoming white, it is not yet white (263b27; b3o). This way of
80 I-=RICHARD SORABJI
putting things seems to neglect the idea that there will be a

stage of being partly white. How then can the transition from
one shade to the next be represented as gradual?
A second way of arguing that the transition is gradual
serves to refute the idea that part by part changing is indis-
pensible for this purpose. The second way is suggested by what
Aristotle says in De Sensu 6. Admittedly, there is only a finite
number of discriminable shades, so that discriminable colours
form a discontinuous series (445b21-9; 446a16-2o). But
nonetheless colours, musical pitches, and other ranges of sens-
ible qualities have a kind of derivative continuity (445b28;
b3o, to me kath' hauto suneches). What Aristotle seems to have
in mind is that a change to the next discriminable pitch, in the
discontinuous series of discriminable pitches, may be pro-
duced by a continuous movement of a stopper along a vibrat-
ing string. Or in the case of colour, a change to the next dis-
criminable shade, in the discontinuous series of discriminable
shades, may be produced by a continuous change in the pro-
portions of earth, air, fire and water in a body. As the stopper
moves along the vibrating string, we hear the sound all the
time, but we do not hear a change of pitch, until the stopper
has moved the distance that corresponds to a quarter tone
(446a1-4). Variations of pitch less than a quarter tone are not
perceptible except by being part of the whole variation
(446a18, hoti en toi holoi), by which Aristotle probably means
that they only contribute to the perceptibility of the whole
variation. This suggests a way in which Aristotle can maintain
that a change to the next discriminable colour or pitch can
be continuous. A body is changing to the next discriminable
shade all the time that the continuous change in its elemental
ingredients is going on, which will eventually lead to its dis-
playing that next discriminable shade.
In what follows, we shall only consider Aristotle's treatment
of the four genuine kinds of change. This will leave open how
he might have treated the many other cases of (non-genuine)
change. In connexion with the four genuine kinds, and the
transitions involved in them, we find two rather different
kinds of treatment.
Aristotle's first solution

The better known one is less satisfactory.It is most fully ex-
pressed in connexion with rest and motion in Phys VI 3
(234a24-bg9)and 8 (239alo-b4). Here Aristotle says that there
can be neither rest nor motion at an instant. He explicitly
cites our problem as one ground for his conclusion (VI 3,
234a34-b5), saying that if something stops moving, the instant
of transition between motion and rest is an instant neither of
motion nor of rest, since otherwise we could not avoid the
contradiction of saying that it is an instant of both motion and
rest. Plato had already given the same solution in the
Parmenides (156C-157A), when he discussed this very puzzle.
At VI 6, 237a14-15, Aristotle extends his treatment of motion
to all change, saying that a thing cannot be changing
(metaballein) at an instant. An extension is also attempted in
VI 8, 239a3-6, where Aristotle denies that there is a first
instant of slowing to a halt, by arguing that slowing to a halt
implies moving, and that there is no moving at an instant.
(To make the argument valid, he ought to show that slowing
to a halt at an instant would imply not just moving, but
moving at an instant.)
Aristotle gives several grounds for denying rest or motion
at an instant, besides the need to avoid difficulties about the
instant of transition between rest and motion. In VI 3
(234a24-b9), one argument is that to rest is to be in the same
state now as then, but an instant does not contain a then.
Another argument is that variation of speed would be impos-
sible at an instant, because such variation would imply that
the faster body had traversed in less than an instant what the
slower body traversedin an instant. Finally, if we cannot speak
of motion at an instant, we cannot speak of rest at an instant,
since we can only talk of rest where there would have been the
possibility of motion.
Aristotle allows that, when something stops moving, there
is a single instant which is both the last of the period during
which the object is moving, and the first of the period during
which it is resting (VI 3, 234a34-b5). And something parallel
is true when a thing starts moving. But this does not in the
least commit him, as he makes very clear, to saying that this
is an instant at which the object is moving or resting.
Since Aristotle thinks his view holds not only for motion
and rest, but for change and stability in general, we can apply
his remarks, for example, to a change of colour, in which a
surface starts off wholly of one shade, and by a gradual process
of transition, finishes up wholly of another shade. If we raise
problems about the first and last instants of its changing
colour, Aristotle will say, for reasons similar to those already
quoted from VI 3, that there is no first or last instant, nor
indeed any instant, at which it is changing colour, or remain-
ing the same colour.
Aristotle sees, however, that this solution is not a complete
one. For although he denies that things can change or remain
in the same state at an instant, he concedes that there are many
other things that can be true of them at an instant. He is
quite prepared to allow that what is moving can be at a point
(VIII 8, 262a3o; b2o), or level with something (VI 8,
239a35-b3) at an instant. As regards other kinds of change, the
object that is changing colour can be white at an instant
(VIII 8, 263b2o; 23), or the white have perished and non-
white have come into being at an instant (263b22). In general,
a change can have been completed, and the new state of affairs
can have come into being at an instant (VI 5, 235b32-236a7;
VI 6, 237a14-15). In allowing something to be white at an
instant, he is not allowing that it could remain white, or rest
in the white state, at an instant.7Since he allows something to
be of a certain colour at an instant, he cannot finally dispose
of our problem by ruling out of order questions about a first
or last instant at which a surface is changing colour. For this
still leaves us free to ask about a first instant at which the
surface is no longer grey, or wholly grey, and a last instant at
which it is not yet white, or wholly white. Aristotle recognizes
the need to deal separately with this further question, and this
brings us on to the second kind of treatment that we find in
his work.
Aristotle's second solution

In Phys VIII 8, 263b15-264a6, Aristotle discusses a change
from not-white to white (or vice versa), and a change from not
existing to existing. He thinks of the final state (e.g., white) as
being reached by a gradual process of transition, but this is
one of the passages where he does not construe the process as

one of white spreading part by part over the surface. Instead,
he implies that throughout the process of transition the surface
will remain non-white (263b27; b30). He distinguishes an
earlier state, by which he means the state when the surface is
still not-white but is changing to white, from a later state, by
which he means the final state of being white. Or rather, since
he switches his example in mid-discussion, the earlier state
is one of being white while changing to not-white, and the
later state is one of being not-white, but for simplicity I shall
stick to the one example. He then says that there is no last
instant of being in the earlier state, but there is a first instant
of being in the later. I take it that it is crucial to understand-
ing the passage to notice that the earlier state, of which there
is no last instant, is one which involves changing' while the
later state does not. His view is generalized in an earlier
chapter (VI 5, 235b6-32),' where it is said in connexion with
all genuine change that there is a first instant of being in the
terminal state after a process of transition. One ground
Aristotle gives for his verdict is again the existence of the very
problem that interests us. He says that the verdict provides a
way (he fails to consider whether it is the only way) of avoiding
the contradiction of something being in its old state and in
its new state at the same instant (263b11; bl7-21). He con-
cedes that there is an instant which is equally the end of the
period during which white was coming into being and the
beginning of the period during which the surface is white, but
he insists that at that instant the surface is already in its later
state, white (263b9-15; 264a2-3).
Aristotle's treatment of the problem here is by and large
very much in line with the solution which we have advocated.
For we should agree with him that there is no last instant of
being not yet white, if changing to white is a continuous pro-
cess. However, we must to some extent qualify our claim to be
in agreement. For when Aristotle gives his reasonsfor denying
a last instant of the earlier state, he does not give our reason,
the continuity of the process of becoming white. This con-
tinuity may well be what influenced him, but if so, he has not
managed to identify it explicitly as the reason."
There are further passages, besides the pair we have men-
4 I-RICHARD S()RABJT
tioned,whereAristotleis attractedto a view close to our own.

We haveso far consideredhis denial of a last instantat which
somethingis non-whitewhile becomingwhite. But he alsodis-
cusseswhetherthere can be a first instantat which something
is changing.At VI , 236a7-27,he wantsto show that there is
no earliest time at which somethingwas changing (236alS,
reading: meteballen),whetherthat earliesttime is construed
as a divisible period,or as indivisible.If indivisible,it might
be construedeither as an instant, i.e. as a boundarywith no
duration,or as an atomof time with an indivisible duration.
In 236al7-20 (whatevermay be true of al6-17), he is con-
struing the putative earliest time as a durationlessinstant.
And he takesthe view whichwe have advocatedin connexion
with continuous changes,that there is not a first instant at
which somethingis changing.Moreover,for the first time he
actuallycontradictshis othersolutionby sayingthat thereis a
last instant at xNhichsomething is resting (contraryto the
doctrineof vr 3 and 8, which denies rest at an instant).His
ground for denying a first instant at which it is changingis
yet again the existence of the kind of problem we are dis-
cussing.Once we assumethat thereis a last instantat which it
is resting,there cannotbe a firstinstantllat which it is chang-
ing. For at such an instant, the object would alreadyhave
changedto some extent,l2and it cannothave changedto any
extent at the veryinstant at which it is still resting. Aristotle
doesnot give as his groundfor denyinga firstinstantat which
somethingis changing the view he takes in VI 3 and 8 that
thereare no instantsat which somethingis changing.He may
insteadbe inSuencedby our kind of consideration;for just as
we associatedmovingwith being awayfromthe startingpoint,
so he associateschangingwith havingchangedto someextent.
And he may be influencedby the fact that in a continuous
change there is no first instant of having changed to some
extent. But if this is whathasinfluencedhim, he hasagainnot
articulatedthe reason.
There is one more place where Aristotlecomesclose to our
sliew.The themeof PhysicsVI 6 is thatwhathaschangedmust
have been changing earlier (237al7-b22), and what was
changing earlier must before that have accomplishedsome
change (236b32-237al7). so that it has already changed an
infinite (237al 1; a16) number of times, and you will never get
a first in the series of changing and having changed (237b6-7).
This implies that there cannot be a first instant of having to
some extent changed, not, for example, a first instant of having
ceased to be wholly grey, or of having started to be partly
white. This implication, which is admittedly not explicitly
spelled out by Aristotle in so many words, is precisely what we
considered to be true of continuous changes.'3
There are then two strands of thought in Aristotle about
the process of transition involved in the four genuine kinds
of change. Sometimes he argues that a thing cannot be chang-
ing or resting at any instant. But sometimes he argues or
implies, in conformity with our view about continuous
changes, that there cannot be a first instant at which a thing
is changing, nor a first instant at which it has left its initial
state, nor a last instant of not having reached its terminal state.
Unfortunately, he does not seem to have a firm grasp of the
latter point of view. For in Phys. VIII 8, 262a31-b3;
b2 W -263a3, he appears to contradict it, by assuming that when
a moving object reverses direction, there is a first instant of
having left the point of reversal. At least, this is the assump-
tion which he seems to require for his conclusion, which is
that the reversing object must spend a period of time at the
point of reversal. The assumptions seem to be that there is a
first instant of having reached the point of reversal, and a first
instant of having left it, and that these cannot be adjacent or
identical instants, so must be separated by a pause during
which the reversing body rests.
Assessment of Aristotle's first solution

Let us return briefly to the first strandof thought, according to
which a thing cannot be changing or resting at an instant.
This view may have appealed to recent philosophers,'"and
is not to be lightly dismissed, but it does have severe disad-
vantages. For it ignores that it is possible, and very useful, to
give sense to the idea of changing at an instant. It is possible,
so long as we acknowledge that change at an instant is a func-
tion of change over a period. It is useful because the velocity
of a body in a given direction at an instant is one of several
factors from which we can calculate in detail its future
behaviour. Much of modern dynamicsdepends on the possi-

of talkingof accelerationat an instant,whereasAristotle
bility
would rob us of this possibility.
Relevance of our problem to the definition of motion at an
instant
Nonetheless, the taskof definingmotionat an instantis by no
means easy.And the kind of problemwe have been discussing,
aboutlastand firstinstantsof restor motion,gainsimportance
fromthe fact that it needs to be resolved,if we are to obtain
asatisfactorydefinitionof motion at an instant. What defini-
But
tionwill be satisfactorydependsin part on our purposes.
some purposes, our
ifmotion is continuous,then at least for
ought to be de-
discussionsuggeststhat motion at an instant
of motion. For a
finedso as to exclude a first or last instant
start,we maysuggestthatan instantof motionwill be one that
be
fallswithin a periodof motion,while an instantof rest will
onethat falls within or boundsa periodof rest. But this defini-
tionmayneed revisionin the light of otherdifficultexamples,
suchas that of the ball thrownverticallyupwards,and slowing
down until it changes direction at an instant. We found the
reasonto regardthis instantas an instantof rest, whereas
definitionjust proposedwould make it an instant of motion,
But
and may need to be revised,for this and other reasons. our
however the deSnition may eventually be formulated,
discussionsuggeststhat for somepurposesit should be formu-
lated so as to exclude first and last instants of continuous
motion,and so as to avoid,if possible,our problemsabout the of
relation to first and last instants of rest. The necessity
gettingclearaboutthesethingsmaynot alwaysbe appreciated. in
BertrandRussellgives a definitionof motion at a moment
§ 446 of The Principles of Mathematics(London 1903, 2nd
edition 1937)and denies,unlike us, that the instant of transi-
does
tion betweenrestand motioncanbe an instantof rest. He
whether or not it can be an
not, however,make it so clear
^ ^
*
lnstant ot motlon.
Generalassessmentof Aristotle'stosition
I should like to finish with a generalassessmentof Aristotle's
his
treatmentof our problem. He has earned notoriety for
refusal to allow motion at an instant. G. E. L. Owen remarks
that this refusal not only "spoilt his reply to Zeno" on the
paradox of the flying arrow,but also "bedevilled the course of
dynamics".15In more detail, Owen explains: "Unable to talk
of speed at an instant. Aristotle has no room in his system for
any such concept as that of initial velocity or, what is equally
important, of the force required to start a body moving. Since
he cannot recognize a moment in which the body first moves,
his idea of force is restricted to the causing of motions that are
completed in a given period of time. And, since he cannot
consider any motion as caused by an initial application of
force, he does not entertain the Newtonian corollary of this,
that if some force F is sufficient to start a motion, the con-
tinued application of F must produce not just the continuance
of the motion but a constant change in it, namely acceleration.
It is the clumsy tools of Aristotelian dynamics, if I am right,
that mark Zeno's major influence on the mathematics of
science."''16
With the first part of this I entirely agree. Aristotle cannot
accommodate the useful concept of initial velocity, by which
is meant, of course, not some first velocity in the entire motion,
but the first velocity which it is convenient to consider in a
given calculation. But what about the second part? A reader
might take "a moment in which the body first moves" to be
a first instant at which the body is moving. We have argued
that it is precisely Aristotle's merit that he denies that there
is such an instant; I would not regard this denial as a defect.
I would accept, then, some charges against Aristotle, but not
others, and at the same time I would draw attention to two
merits of his discussion. First, it is a merit that he recognizes
that not all cases call for the same treatment. The treatments
we have been discussing apply only to those processes of
gradual transition which he believes to be involved in the four
genuine kinds of change. Second, Aristotle does express the
view which we believe to be reasonable for continuous change,
namely that there is no first instant at which a thing is chang-
ing, or at which it has begun to abandon its original state,
and no last instant at which it has not yet achieved its final
state."
88 ---RICHARDSORABJI
NOTES
1 This revival has been discussed by Curtis Wilson in William Heytesbury,

Madison Wisconsin, 1956, and now in an illuminating paper by Norman
Kretzmann,"Incipit/Desinit" (see n. 18 inf.). This should be read by anyone
interested in the continuing history of the subject, although we diverge in our
interpretationsof Aristotle.
2 It may be objected that if the atoms of a body are for ever joggling, and
if their motions are not equal and opposite so that they cancel each other
out, then the centre of mass will also be for ever moving, so that it will have
no first instant of motion. We may reply that, even if this is so, we can still
ask about the first instant of motion (and last of non-motion) of the centre
of massin a given direction, or in responseto a given force. I shall neglect this
complication in what follows.
3 This consideration would fall to the ground, if on other occasions the
centre of masswere deflected at an angle without the velocity slowing to zero.
For then the positive velocity would persuade us to talk of motion at the
instant of deflection, and we should then be accustomedto talking of motion
without there being one direction rather than another. But if deflection at
an angle (as opposed to deflection in a curve) is impossible (and it would
involve a discontinuity in the velocity in a given direction), then our
considerationcan stand.
4 Accordingto Kretzmannop. cit. It looks as if Peter of Spain failed to see
that whether one's interest is in identifying the instant as it arrives is quite
independent of whether one is discussing what he calls "permanent" states.
or "successive"states, or the beginnings or endings of either.
s In effect, he defines, "it was in motion (not at rest) at instant t," by
saying something like, "t was either followed, or preceded, or both, by a
period throughout which it moved." And he defines, "it was at rest (not in
motion) at instant t", roughly as "t waseither followed, or preceded, or both,
by a period throughout which it did not move".
6 Aristotle probablyhas this answer in mind also in VI 5, 236b5-8, where he
stresses that, even if colour is not in itself divisible, the surface to which it
attaches is divisible.
?All he is committed to is the view expressed elsewhere (Phys VIII 8,
264b1), that what is white must remain white over a period.
8 That he thinks the
process of change is going on is clear e.g. from
263b21-2, "non-white was coming into being, and white was ceasing to be".
263b26-7, "If what exists now, having been previously non-existent, must
have been coming into being, and did not exist while it was coming into
being . . .". 264a2, "The time in which it was coming to be". 264a5, "It
was coming to be".
9 There is such a thing as the time "when first a thing has changed"
(VI 5 235b7-8; b31; b32), i.e. has completed its change. This is an indivisible
time (235b32-236a7).And at that instant the thing is already in its new state
(235b8; b31-2).
10Aristotle's solution is only appropriate, given his two assumptions that
the discriminable shades form a discontinuous series, and that nonetheless
the change from one discriminable shade to the next is a continuous one. A
quite different treatment of colour changes would be called for, if he took
the view (i) that colours form a continuous series. In that case, the possibility
would arise of producing a continuous alteration of shade along the spectrum.
In a continuous change of shade such as this, there could not be a last instant
of one shade, nor a first of another. The situation would be different again,
if (ii) he took colours as forming a discontinuous series, and also took the
change from one colour to the next as being discontinuous. In that case,
nothing we have so far said would enable us to decide which colour to speak
of at the instant of transition from one colour to the next. As fox (iii) a
discontinuous transition through several intervening shades to a distant one,
such a transition might be thought of (though it need not be) as having a
first instant (namely, the first instant at which a colour other than the original
one existed), and a last instant (namely, the last instant at which the
penultimate colour was in existence).
11It is confusing that this putative instant is referred to by two letters,
A A, and not just one. The reason is that A A stands for the putative earliest
time of changing, which is later treated as a period with A and A as its
terminal instants. Here, however, it is treated as a durationless instant, so
that & is not separate from A.
12 Or "have begun to change". At 236a7-lo "has changed" (metabeblike)
is said to be ambiguous between "has completed its change" and "has begun
to change". The latter sense is relevant in the present lines (236a19-2o); the
former is not.
13 There is another way in which this passage diverges from Aristotle's first
solution. Though there is no first instant of having to some extent changed,
there is an instant which divides the period of stability from the period of
change. This instant is the last instant of the period during which the object
is not changing, and Aristotle's view elsewhere (234a34-b5; 263b9-15) suggests
that it can also be called the first instant of the period during which the
object is changing. But at one point in the present chapter, VI 6, wittingly or
not, he casts doubt on the latter description. For he says (237a15) that at any
instant of the period during which a thing is changing, it has already changed.
This would seem to rule out not only a first instant of having to some extent
changed, but also the applicability of the description "first instant of the
period during which a thing is changing." Aristotle thereby contradicts for a
second time an aspect of his other solution.
14 I am not sure whether this is the intention of Vere
Chappell in "Time
and Zeno's Arrow", Journal of Philosophy 1962.
15 "The Platonism of Aristotle", Proceedings of the British Academy 1965,
p. 148. Similarly "Aristotle" pp. .25-6 in vol. I of The Dictionary of Scientific
Biography, ed. C. C. Gillispie, Scribners, N.Y. Owen's papers on the
continuum are required reading for students of this subject.
16 "Zeno and the Mathematicians", Proceedings of the Aristotelian Society.
1957-8, pp. 220-2.
17 I have had
many helpful discussions on this topic, but I am particularly
indebted to Geoffrey Lloyd for a thorough correspondence about some of the
texts, to my student Marcus Cohen and to Malcolm Schofield for some very
helpful discussions on the philosophical issues, and to Professor Clive
Kilmister for patient advice on Newtonian mechanics.
18In Machamer
&8Turnbull edd., Motion and Time, Space and Matter
(Columbus: Ohio State U.P., 1976), pp. 101-136.
ARISTOTLE ON THE
INSTANT OF CHANGE
Richard Sorabji and Norman Kretzmann
II--NormanKretzmann
TIME EXISTS-BUT HARDLY, OR OBSCURELY
(Physics IV, 10o;217b29-218a33)

It is characteristic of Aristotle to preface the explicit presenta-
tion of his own view with reviews of difficulties proper to the
subject matter and surveys of his predecessors' opinions. He
follows that practice in his treatise on time (Physics IV, 10-14),
beginning with a clearly delineated investigation of four
puzzles (dieporemena) arranged in two pairs.' Regarding the
puzzles generally he says that it is good to begin by considering
them (217b30), that they are associated with popular or
familiar views of time (dia ton exoterikon logon-2 17b30-3 1),
and that they, like the opinions of his predecessors,fail to re-
veal the nature of time (218a31-33). The puzzles have been
taken to be unresolved problems which Aristotle afterwards
ignores.3 I think, on the contrary, that they are, or involve,
arguments designed to show that the most natural and
familiar view of the nature of time is incoherent. They fail to
reveal the nature of time, but they show what time cannot be.
And if they do what I think they do, the four puzzles make
a very important contribution to the understanding of
Aristotle's own theory of time.
The Four Puzzles

I. Time consists entirely of two nonexistent parts; and
II. The present is not a part of time.
The present instant (the now) is either (III) always
different or (IV) always the same; but
III. If the now is always different, it is so either because
IIIa. there are different simultaneous nows-which
is impossible-or because
92 II-NORMAN KRETZMANN
IIIb. there are differentsuccessivenows-which is

impossible-and
IV. If the now is alwaysthe same,all events are simulta-
neous-which is impossible.
After a brief introductoryparagraphthe puzzles passage
gets under way with the statementsof two counter-intuitive,
or at least disconcerting,claimsregardingthe realityof time:
Time is altogethernonexistent;Time exists-but hardly,or
obscurely.These claims will emergeas forced conclusionsat
differentpoints in the developmentof the puzzles;and if I
am right about the aim of the puzzles,the conclusionswill
constitute embarrassments for the familiar view that time is
essentially, really passing.
I
The development of the first puzzle may be conveniently
presented in the following close paraphrase.
Time-either infinite time itself or any time one might
choose-is composed of two parts. One of those two parts
has come to be and is no longer; the other part will be and
and is not yet. Therefore neither part of time is. And
therefore time is altogether nonexistent, for what is
composed of nonexistent parts does not itself exist.
There are two obvious objections to this argument. In the
first place, it seems plainly false that any time one might
choose is divisible into the two parts described in the argu-
ment. Suppose we choose the nineteenth century. We might be
willing to say that it, like time itself, is composed of a past
and a future; pick a date, any nineteenth-century date, and it
will mark off a nineteenth-century past and a nineteenth-
century future. We are sometimes willing to extend the use
of the words 'past' and 'future' in that way; but Aristotle does
not use the Greek words for past and future here. He says of
time itself that one part of it has come to be and is no longer
and the other part will be and is not yet, and we cannot say
that of any time we might choose-of the nineteenth century,
for example, or the twenty-first. We can say that only of any
time we might choose which, like infinite time itself, includes
now (but not as either of the limits of the time); suppose we

call such a period of time an "N-period". Any and every time
is divisible into parts, but only an N-period, no matter how
long or how short it may be, is divisible into one part that is no
longer and another part that is not yet. Must we, then, con-
sider the first puzzle as applicable only to N-periods? I think
we must.' But, of course, any time we might choose which is
not an N-period is a time which as a whole either is no longer
or is not yet. It is therefore only N-periods that have any chance
of counting as existent or real, even if only in part; and so the
fact that the first puzzle applies only to them is a mark of
economy, not of weakness.5
The second of the two objections is even more obvious than
the first. As Fred Miller puts it, "One might be willing to
concede the point about the unreality of past and future, and
yet reject the conclusion on the grounds that the argument
has overlooked a very important part of time: the present".6
We may assimilate this second objection to the first by think-
ing of it as the claim that an N-period is not exhaustively
divisible into a part that is no longer and a part that is not yet.
I think Aristotle expects and intends his readers to raise this
objection; it sets them up for the second puzzle.
The complementary relation of the second puzzle to the
first is indicated also by the fact that the second is prefaced
by a restatement of the first, as if to drive home the pre-
liminary point.
In addition, if any thing with parts is to exist, then, when
it exists, all or some of its parts must exist. But, although
time is divisible [i.e., although time has parts], some parts
of it have been and the others will be, and no part of it
exists (218a3-6).
In the original statement of the first puzzle Aristotle spoke of
two parts of time--one that has been and one that will be. In
this restatement of it he speaks, in a way that at first seems
unnatural, of parts that have been and other parts that will be.
But these plurals are well suited to the considerations raised in
connexion with the introduction of the notion of an N-period.
All parts, or periods, of time are either N-periods or not. If
they are not N-periods, then they are parts that have been or
94 II--NORMAN KRETZMANN
parts that will be. If they are N-periods, then they have parts
that have been and parts that will be.
Before considering the second puzzle directly, it is worth
noticing that the language of the first puzzle, both in the
original statement and in the restatement, leads naturally to
the inference that infinite time itself or any finite N-period
is not divisible without remainder into the parts Aristotle
describes. In the original statement the one part is no longer7
and the other part "is not yet", and in both the original and
the restatement he says of the parts that they have been (or
have come to be) and will be. Such expressions cannot apply
unless regarding the one part (or each thing in the one part)
it has been the case that it is, and regarding the other part
(or each thing in the other part) it will be the case that it is. So
if time is continuous-and Aristotle of course maintains that
it is so (219a12-13; 233alo)-there must be something
temporal that is. And so it must not be true that time is ex-
haustively divisible into what has been and what will be;
time must, as the language of the first puzzle itself implies,
include what is. But what can that be? It cannot be an
N-period, for such a period, like infinite time itself, also in-
cludes what has been and what will be. Yes, but even when
they are stripped away,8there is a remainder: now, the present
instant'-which must be the part of time that is.
II
That elaboration of the second objection to the first puzzle
leads directly to the second puzzle, the heart of which is the
flat denial with which the presentation of it begins: "as for
the now, it is no part of time" (218a6). What lies behind the
denial, of course, is the recognition that the now is only the
division between the parts, just as the twenty-four hour day
is exhaustively divided into A.M. and P.M., 12: oo M. being
only the division between those parts. The now is an instant,
the temporal analogue of a geometrical point.
But someone who objects to the first puzzle on the grounds
that it leaves the present out of account might easily fail to
appreciate the force of the second puzzle against his position,
even if he agrees that the present must be conceived of as a

bare instant. Even if he knows that a point has no extension
and agrees that an instant is like a point in having no duration,
the sophisticated objector may seek further justification for
Aristotle's announcement in the succeeding chapter that an
instant is no more a part of time than a point is a part of a
line (22oa19-20). Why, he will want to know, should he not
say of a bisected line segment that it has three parts-the left
half, the right half, and the bisector (or point of bisection)?
And why, even though an instant has no duration, should he
not say of time that it has three parts--one time that is no
longer, one time that is not yet, and the instant that is now?
Aristotle's answer is contained in the development of the
second puzzle, but not, I think, very clearly.
And as for the now, it is no part of time; for a part
measures, and the whole must be composed of the parts,
but it is not thought that time is composed of instants
(216a6-8).
We seem to have an argument of three premisses:
(1) a part measures,
(2) the whole must be composed of the parts,
(3) time is not composed of instants."
We may take it for granted that we are dealing here exclusively
with merely quantitative wholes, such as an hour, time, a line
segment, space. The strategy of the argument is to lay down
criteria for the parts of a merely quantitative whole and to
point out that the now (or an instant generally) fails to meet
those criteria. If I have understood the argument correctly,
Aristotle's intentions in it might be more clearly represented
in this way:
(i) A part must measure the whole

(ii) A part must make a quantitative contribution to
the whole
(iii) An instant neither measures nor makes a quantita-
tive contribution to time
.(iv) An instant generally, or the now in particular, is
no part of time.
The differencesseparating(i) and (ii) from (1) and (2) are

slight enough, perhaps,to be consideredonly accidents of
diction. (iii) is more substantiallydifferent from (3), but
without the addition effected in (iii) I cannot see how the
argumenthangs together.With these revisionsthe argument
does its job fairlywell, I think, althoughI have somereserva-
tions about the firstpremiss.
The firstpremissmustmeansomethinglike this: Whatever
countsas a part of somewhole is somethingin termsof which
that wholemaybe measured.By this criterionan inch maybe
countedas a part of a mile or of a journey,and a secondas a
part of a day or of a lifetime. And by this criterion,it may be
acknowledgedat once, an extensionlesspoint does not count
as a part of a line, and a durationlessinstantdoes not count as
a part of time. But whatever might be said in general on
behalf of the view of the nature of a part on which this
criterion is based, it seems peculiarlyinappropriatein this
context. Any relevant denial that the now is a part of time
must contrastthe now appropriatelywith the past and the
future,which have alreadybeen described(in puzzleI) as the
partsof time or of any finite N-period;and the past and the
future areno bettersuited to serveas measuresof time than is
an instant. Past and future are not partsof time in the way
in which secondsor centuriesare,and yet thereseemsto be no
good reasonto object to their designation,in puzzle I or else-
where, as partsof time. Parts of the sort that can be used as
measures-mensurantparts,we maycall them-must (a)have
somefinite quantityand (b) be quantitativelystable.1
As partsof somefinite N-period-to-day,for example-past
and future do have finite quantity, but as parts of a finite
N-periodtheyare systematicallyunstablequantitatively;their
quantities are incessantlychanging relative to each other.
Supposesomeoneis now drivingfrom Ithacato New Yorkvia
Binghamton.We might measure the length of the journey
in miles, of course;we might also measureit in lengths of the
carbeing driven-so manyVWs.We might measurethe time
of the journey in hours, of course;we might also measureit
in times of a particularpart of the journey-so many Ithaca-
Binghamtons.But we could not measurethe distancein terms
of the distance covered as of now or the time elapsed as of

now-and I mean as of now, not as of whatever clock-time it
happens to be on the occasion of one of my utterances of 'now'.
Thus although every N-period is composed of past and future,
past and future are not mensurant parts of a finite N-period.
As parts of infinite time itself, past and future are both in-
finite; they thus lack the first necessary characteristic of
mensurant parts--finite quantity. And although they are both
(always) infinite, they may be said to be quantitatively un-
stable as well, since, for instance, today's past cannot be
mapped onto yesterday'spast without remainder.
The first criterion thus fails to contrast the now with the
past and the future in the relevant respect. To observe that a
part measures the whole would effect the required contrast
only if the now were being compared with some genuinely
mensurant part, such as a second. The now, like any other
instant, is quantitatively stable-it never varies from zero
quantity--but it thus fails to satisfy the other necessary con-
dition of mensurant parts. The now is not a mensurant part
of time; but neither are the past and the future. I suspect that
Aristotle employed this first criterion casually, realizing that
only a positive quantity could serve as a measure but neglect-
ing to take into account the complicating factors I have been
pointing out regarding mensurant parts.
Even if I am right in thinking that the first criterion in the
argument intended to develop the second puzzle is inappro-
priate, it does, after all, support the conclusion: the now is not
a (mensurant) part of time. And in any case the second
criterion upholds the conclusion in an appropriate way.
If we think of to-day as a salami sausage, then the present
instant is a cut dividing the sausage in two. While the salami
can be reconstructed out of slices, no matter how thin, it
cannot be reconstructed out of cuts, no matter how many. As
a salami is not composed of cuts, so a day (or infinite time
itself) is not composed of instants. If it is right to say without
qualification of any whole that it is the sum of its parts, it must
be right to say so of a merely quantitative whole. Whatever
counts as a part of a whole which is the sum of its parts is
a summand, makes a quantitative contribution to the whole.
98 I--1NORMAN KRETZMANN
Past and future are summand parts of time; and now is not.
Considering the now a part of time is like considering the part
in your hair a part of your head.
I 8cII
Before considering the second pair of puzzles I want to spell
out the position on the reality of time which seems to result
from the first pair of puzzles together with a few fundamental
Aristotelian views on the nature of time. We have been pro-
ceeding so far, following Aristotle's lead, on the familiar view
that time consists entirely of the past, the now, and the future,
which may be characterizedalso as the view that time is essen-
tially passing. On that view the past and the future are the two
parts of time, distinguished from each other by the now, which
is of course temporal, but not a part of time. The past is not;
the future is not. It follows that no part of time is. Ordinarily,
if neither of two parts exists, then whatever has been divided
exhaustively into those two parts does not exist. One might
therefore feel justified, if uncomfortable, at this stage in say-
ing that time does not exist. The uncomfortable feeling should
stem not only from the bizarre character of the conclusion but
also from the still shadowy status of the now. Although it is
not a part of time, it is temporal, and it is. The now is an
instant, and "an instant is not time but an attribute" of time
(22oa2 1-22). But an instant must be an attribute of time in
the essence-delineating sense in which mortal rational
animality is an attribute of man, for "it is also evident that
neither would an instant exist if time did not exist, nor would
time exist if an instant did not exist" (2 19b33-220a1). And
since Aristotle certainly does speak as if instants exist, it
looks as if it must be in virtue of the existence of an instant
(of the now in particular) that he is able to say that "time
exists" (222b27). But when he comes to state his own theory of
time he says (in the succeeding chapter) that "time does not
exist without a motion" (21 8b33). And in leading up to his
discussion of Zeno's paradoxes of motion (in Book VI) he says
that "a thing can be neither at rest nor in motion at an
instant" (239b1-2). So if there is no motion at an instant, and
no time without motion, there is no time at an instant. That
conclusion may seem to be a truism; of course there is no time

at an instant-no more than there is space at a point. But if
there is no time at an instant, and if an instant is no part of
time, and if the two parts of time do not exist, and if "an
instant is not time" (22oa1), then even if we acknowledge that
an instant-the now-exists, how does that acknowledgement
warrant our concluding that time exists? Well, Aristotle says
that the now is an attribute of time, and if an attribute exists,
then surely its substrate exists. But where is the time that is to
serve as the substrate for that attribute? There simply is none
available-not, at any rate, as long as we proceed on the
familiar view that time consists entirely in the past, the now,
and the future. Moreover,it is, after all, the now which defines
the past-before now-and the future-after now. If the
now is to be describedas not time but an attributeof time, it
seemsthat the past and the future are to be describedin just
that way as well. On the basis of this review of the first pair
of puzzlesit seemsquite right to suggestthat theymight lead
one to suspectthat "time is either altogethernonexistent,or
that it exists,but hardly,or obscurely".
III &cIV
The two puzzles constituting the second pair occur as the
horns of a dilemma. Either "the now which appears to divide
the past and the future [IV] alwaysremains one and the same,
or [III] it is always distinct" (218a8-io). It seems clear that the
theoretically possible middle ground-that the now some-
times remains one and the same but occasionally changes-
is ignored as unthinkable. The question Aristotle is consider-
ing here resembles the question whether the sun that rises in
the morning is always the same or always a new sun, and to
that question the middle-ground answer is not unthinkable.
But in the case of this question about the now we do not have
lots of similar appearancesof some entity or entities, as in the
case of the question about the rising sun. Here we are seeking
a characterizationof "the now which appearsto divide the past
and the future", which is paradigmatically always; it is always
now. We seem, therefore, to be offered a forced choice between
the only two thinkable accounts of what is always going on if
anything at all is going on.
100 I.---NORMAN KRETZMANN
The choice is an odd one, perhaps odder than it looks at

first. For I think we are likely to feel we recognize the first
of the two options; it seems to be the account of the now which
forms a part of the familiar view of the nature of time-the
doctrine of the nunc fluens. "We are inclined to think of the
now as surging through history like the crest of a wave along
the ocean surface-or 'we think of the present as a spotlight
that plays along a line of chorus-girl-like events' ".* These
pictures, designed to present the familiar view, certainly seem
to be portraying a now that "always remains one and the
same". As I shall try to show, however, the association of the
first option (puzzle IV) with the nunc fluens doctrine is
mistaken.
III
In his discussion of the dilemma Aristotle takes up the second
option first; it thereby becomes the third of the four puzzles
to be developed. This third puzzle deals with an account of
the now which seems unfamiliar and difficult to characterize
precisely; it may be tentatively designated the nunc differens
account. Any attempt to characterize it further is best post-
poned until we have seen how Aristotle handles it. His discus-
sion of it is complicated; perhaps a simple preliminary obser-
vation will be helpful. The claim is that the now is always
distinct, but there are, one might suppose, two ways in which
there might be distinct nows-simultaneously or successively.
Thus the development of puzzle III, itself an option in a
dilemma, takes the form of a dilemma. The long opening
sentence of the discussion (218al1-16) aims at dismissing the
possibility of simultaneous distinct nows; the remainder of the
paragraph (218a16-21) then attacks the possibility of succes-
sive distinct nows.
IIIa
Most of us are unlikely to feel initially baffled by the notion of
successive distinct nows, but what might lead one even to
think of simultaneous distinct nows? In the sentence in which
he attacks the notion of simultaneous distinct nows Aristotle
speaks of "distinct part of time" (218a11-12), and if the parts
of time are cut up in a certain (quite natural) way, one might
be led to think casuallyin terms of simultaneous distinct nows.

Take, for example, my lifetime and the lifetime of my cat
Suki, who is now dead. The two lifetimes undoubtedly are
parts of time (indeed, mensurant parts of time), and they are
simultaneous in the respect Aristotle specifies: Suki's lifetime
is contained by my lifetime; "..a. no two distinct parts of time
exist simultaneously unless one part contains while the other
is contained, as the smaller period of time is contained by the
greater" (!218al -14). If Suki's lifetime and mine will serve
as simultaneous distinct parts of time, it does not seem un-
natural to think of them as parallel times, so that for every
instant of Suki's lifetime there is a corresponding instant of
mine-simultaneous S-instants and NK-instants. The induce-
ment to think in terms of simultaneous distinct nows becomes
even stronger if we change the example from me and my cat
to you and your absent friend; it may seem entirely natural to
think of his or her now as distinct from yours. But the
apparent reasonableness that can be built up in the notion of
simultaneous distinct nows is, I think, all a consequence of
misleadingly specific descriptions of parts of time, such as my
or your, his or her, lifetime."3If Aristotle had meant to be
considering parts of time distinct in that highly specified way,
then he ought to have admitted complete coincidence of two
times (or parts of time) as well as containment of one by the
other. I think Aristotle's omission of complete coincidence as
a case of simultaneity of two distinct parts of time is deliber-
ate. If there is complete coincidence, then there are not two
distinct parts of time but only two distinct descriptions of one
part of time. If the lifetimes of Robinson and Carpenter are
completely coincident, they no more constitute distinct parts
of time than do the lifetime of Robinson and the calendar
time marked off by the instants of his birth and his death, or
the time during which I look at you and the time during
which you are looked at by me.
According to Aristotle's criterion the cat's lifetime and mine
do qualify as two distinct, simultaneous parts of time, but be-
cause we are likely to be misled by a consideration of such
highly specified parts of time, we will do better to consider
merely clock-times. The four-hour period from eight until
noon to-day is a part of time, and the three-hour period from
nine until noon to-day is a part distinct from but simultaneous

with the four-hour period. No one is tempted to say that an
instant of the shorter period has a corresponding, simultaneous
instant in the longer period; the only sensible thing to say
is that every instant of the shorter period is an instant of the
longer period.'4By the way in which he treats the possibility
of distinct and simultaneous parts of time Aristotle shows that
he takes time to be linear. On that view of time-so natural
to us that it scarcely qualifies as a view-there can no more
be distinct but simultaneous instants than there can be
distinct but coincident points on a line.'5
But in the complicated conditional sentence we are con-
sidering Aristotle does not go directly from the observations
about simultaneous parts of time to the consequent "then two
instants cannot exist simultaneously" (218a 15).16 He first intro-
duces an additional consideration-to de nun me on proteron
de on ananke ephtharthai pote (218a14)-which is open to two
interpretations, depending on whether 'nun' is taken as a sub-
stantive ('the now' or 'an instant') or as an adverb ('now')
contrasting with 'proteron' ('earlier'). The first interpretation
may be conveniently designated 'W-A' after Wicksteed and
Apostle, who use it in their translations," the second, 'C-R',
after Cornford and Ross."
W-A: an instant which does not exist but existed earlier
must at some time have been destroyed
C-R: that which does not exist now but existed earlier
must at some time have been destroyed.
W-A is supported by the context, which is studded with
references to instants or the now, and by the fact that the ad-
verbial use of 'now' seems not to add anything to 'does not
exist'. On W-A, however, Aristotle's additional consideration
regarding instants seems, although obviously true, irrelevant
to his attempt to show that instants cannot be simultaneous.
Without this passage the only explicit reason against the pos-
sibility of simultaneous instants is the observation that distinct
parts of time are simultaneous just in case, and only to the
extent to which, one part contains (or overlaps) the other.
Even though Aristotle has already said that instants are not
parts, it would be helpful, especially at this very early stage of
his discussion of time, to say expressly that the sort of simul-

taneity available for distinct parts is unavailable for distinct
instants, to point out that there is no possibility that some now
of the very recent past might last long enough to contain (or
overlap) this now. For instants, of course, differ from parts of
time in that respect. A part of time that has been need not be
a part of time that is not, that must already have ceased to be-
take to-day,for example, or any finite N-period. And Aristotle
could have made this point by inserting an additional con-
sideration which bears a strong resemblance to W-A:
an instant which existed earlier does not exist but must at
some time have been destroyed.
But the strong resemblance is superficial; the Greek as it
stands will not bear this useful interpretation. If we take the
natural course of reading 'nun' as a substantive, we are stuck
with the pointlessnessof W-A.
As for C-R, it has one obvious advantage over W-A (and,
for that matter, over the useful but insupportable interpreta-
tion) in that it leaves a point to be made in the concluding
clause of the sentence-ephtharthai de ananke aei to proteron
(218ai 5-16) ("but always the earlier instant must have
been destroyed")-which on W-A looks like mere repetition.
Moreover, on C-R line 14 loses the obvious pointlessness
it has on W-A; but it does so, I think, only because its point-
lessness has been obscured a little. Instead of being a truism
regarding instants in particular it becomes a general truism.
Following C-R, the argument against the possibility of simul-
taneous distinct nows (the development of puzzle IIIa) may be
paraphrasedin this way:
Suppose that the now is alwaysdistinct; then since no two
distinct parts of time exist simultaneously unless one part
contains the other, and since that which does not exist
now but existed earlier must at some time have been
destroyed, two instants cannot exist simultaneously, but
always the earlier instant must have been destroyed.
But this argument is invalid. The useful interpretation of line
14 (the second premiss) would validate the argument, and it is
tantalizingly close to the actual text, but I am not bold enough

to suggest the emendation that would bring it within reach.
Nuances and conflicting interpretations aside, the intended
achievement of the complicated conditional sentence is clear
enough and may be granted (even if not on the grounds
provided in the sentence itself): if the now is always distinct,
it cannot be so in virtue of simultaneous distinct nows. If my
general outline of the paragraph is correct, it remains to con-
sider the second option of the sub-dilemma (the second half
of the third puzzle), the possibility that the now is always
distinct in virtue of successive distinct nows.
IIIb
The consideration of that possibility begins in a way which
shows that the final clause of the conditional sentence-"but
always the earlier instant must have been destroyed"-is
doing duty also as an introduction to the next stage of the dis-
cussion; for the development of the second half of the third
puzzle may be said to begin (implicitly) in this way: If there
are successive distinct nows, then each now that has been must
have ceased to be. And the development proceeds in an
attempt to show that the notion of an instant's ceasing to be is
incoherent. I want to paraphrase that attempt before looking
at it more closely.
Let t, be some instant that has existed and does not exist.
Then tP must, at some instant, have ceased to exist. Let tn
be the present instant (now). Then, since tp does not exist,
at
tp must at some past instant have ceased to exist-either
t, itself or at some instant between tp and tn. But tp cannot
have ceased to exist at t,, for that is just exactly when it did
exist. So it must be that tp ceased to exist either at the instant
immediately after t, or at some subsequent instant (before tn).
But there is no instant immediately after tp-"instants, like
points on a line, cannot be consecutive to each other" (218a
18-19).19 So t, must have ceased to exist at some instant
between tP and t, (but not immediate to either t, or tn)-
call that instant tin. Let tm be arbitrarily soon after tp; there
will nevertheless be infinitely many instants between t, and
tin. So if tp ceased to exist at "it would have existed simul-
ti,
ARISTOTLE ON TIlE INSTANT OF CHANGE 105
taneously with the infinitely many instants between itself

and that other, and that is impossible" (218a 19-21). (The
impossibility of simultaneous instants may be taken to have
been shown in the development of puzzle IIIa.)
We are faced with the apparently paradoxical conclusion
that no instant ever ceases to exist, a situation in which we
are bound to review the steps which brought us to such a
pass. There is nothing wrong with the rejection of the possi-
bility that t, ceases to exist at some time between to and t,
or with the denial of the possibility of consecutive instants.
The paradox, if it is a paradox, must stem from the rejection
of the possibility that t, ceases to exist at t,.
Mediaeval logicians dealing with problems associated with
beginning and ceasing sometimes recognized instantaneous
existence-along with permanent and successive existence-
to cover primarily just this case of the instant, which begins
to exist, exists, and ceases to exist simultaneously.? If instants
exist, then surely that is how they exist-instantaneously. As
we have seen, Aristotle maintains that instants do exist,2'but
in the development of this part of the third puzzle (IIIb) he
in effect rejects instantaneous existence as impossible: "the
earlier instant cannot have been destroyed at that very instant
itself, because it existed then" (218ai6-17). The rejection is
certainly plausible; instantaneous existence is at best a prob-
lematic notion. But the rejection is, I think, inconsistent with
Aristotle's own commitment to the existence of instants. One
might suppose that he employs it only for the sake of the de-
velopment of puzzle IIIb. But if the aim of that development
is solely to argue that the now cannot be always distinct in
virtue of successive distinct nows, Aristotle has open to him
a more powerful line of argument which is consistent with
his own position. For he might have argued for that conclu-
sion on the grounds that if "the now which appears to divide
the past and the future" were always distinct in virtue of suc-
cessive distinct nows, there would be one now right after
another,2 which is impossible. His neglect of this line is the
more surprising in view of the fact that he presents the basis
for it in the very development of IIIb: "instants, like points
on a line, cannot be consecutive to each other" (218a18-19).
As in the development of puzzle IIIa, he seems here to have
used a bad or at least a suspect argument where a good one lay

ready to his hand.
I can think of only one explanation for his doing so in
this case, and it is not a very good one. The better argument
would not yield the conclusion that no instant ever ceases to
exist, and that conclusion seems to have a dialectical function
in the structure of the second pair of puzzles. If we have to con-
clude that no instant ever ceases to exist, we are going to feel
even more inclined, though for an unexpected reason, to take
seriously the claim dealt with in the fourth puzzle-the view
that the now "alwaysremains one and the same".
When I began considering puzzle III (on page too above), I
said that we might be in a better position to characterize the
nunc differens account when we had looked at Aristotle's
attack on it. The principal clarifying effect of the attack is that
it leaves us with only the IIIb version of nunc differens to
consider seriously; the IIIa version, of simultaneous distinct
nows, is too obviously farfetched. As for successive distinct
nows, a theory maintaining that the now "is always distinct"
in that respect must maintain that instants cease to be and
that one instant immediately succeeds another; puzzle IIIb
has raised insuperable difficulties for nunc differens on at least
one of those two counts without adding any further details.
The absence of detail coupled with the oddness of the
original difficulty-Does the now which appears to divide the
past and the future always remain one and the same or is it
always distinct?--suggests that it is a mistake to think that
what Aristotle is dealing with in this second pair of puzzles
is two rival accounts of the nature of the now of temporal
passage-accounts broadly characterizable as nunc differens
and nunc fluens. If we do think in those terms, most of us will
feel more at home with the nunc fluens account, the view that
time is composed entirely of the past, the now, and the future,
and that the passage of time is, more precisely, the passage of
the now. But even though the designation "the flowing now"
and the familiar images of the rolling wave and the traveling
spotlight suggest a now that "always remains one and the
same", we cannot take seriously, or imagine Aristotle taking
seriously, the notion that there is a persistent instant flowing
along the time line, thereby creating the past out of the future.
ARISTOTLE ON THE INSTANT OF CHANGE
1o7
Traditional designation and imagery to the contrary notwith-
standing, we come closer to depicting accurately the now of
temporal passage, "the now which appears to divide the past
and the future" if we say that it is "always distinct". Looked
at with some awareness of the nature of instants, the so-called
nunc fluens seems to disintegrate into the nunc differens.
With some hesitation I suggest, then, that what puzzle IIIb
really attacks is not some obscure theory regarding the nature
of the now but the familiar view that time is essentially, really
passing. If time consists, as the first pair of puzzles takes for
granted, entirely of the past, the now, and the future, then
time is essentially the passing of time. And if time exists but
the future and the past do not exist, then the now exists. And
if the now exists and time passes, then the now ceases to exist,
again and again, infinitely often in any temporal interval no
matter how short. But there is an incoherence either in the
notion of a now's ceasing to exist or in the notion of one now's
succeeding another consecutively. Therefore there is an in-
coherence either in the notion that the now exists, or in the
notion that time exists, or in the notion that time is essentially,
really temporal passage. And we know that Aristotle upholds
the notions that the now exists and that time exists.
III to IV
Despite the close connexion between puzzles III and IV as the
two horns of the same dilemma, the transition from III to IV
shifts the focus of attention from temporal dynamics-the
passage of time-to temporal statics-the permanent ordering
of events as earlier (or later) than one another." Aristotle
presents his own definition of time in terms of temporal statics
(as the measurable dimension of motion in respect of earlier
and later (at 219b1-2)) and expressly denies that time itself
is a motion or change (218bi8-20). This is not the occasion on
which to try to expound Aristotle's theory of time, but I can-
not in good conscience avoid saying that I think he takes time
to be essentially the permanent ordering of events and that he
considers the passage of time to be an attribute of, or the
appearance of, that linear sequence.2 The stage is set for his
theory in the discussion of the puzzles, the first three of which
108 KRETZMANN
II---.NORMAN
seem designed to show that one cannot consistently hold that

(A) time is real and (B) the essence of time is temporal passage.
Aristotle himself holds (A), but, I maintain, in place of (B) he
holds that the essence of time is the fixed temporal order. His
discussion of puzzle IV tends to confirm this interpretation
since in it the reality of the temporal order is taken for granted
and serves as the basis for challenging the view that the now
"alwaysremains one and the same".
IV
The fourth puzzle is developed in two arguments. The first
of them, although it is a simple argument based on elementary
facts about the temporal order, may not be quite straight-
forward in its application to the view under consideration.
Moreover, neither is it possible for it to remain always the
same; for no finite and divisible thing, whether continu-
ous in one or in more than one dimension, has just one
limit, but an instant is a limit, and it is possible to cut
off a finite time (218a21-25).
We can pick out a period of time, a period of time must be
limited at both ends, instants are the limits of periods of time;
therefore there cannot be only one instant. That much of the
argument has nothing to do with the passing of time; the con-
clusion as I have drawn it-there cannot be only one instant
-has to be interpreted tenselessly. But the conclusion at
which this argument is ultimately aiming is that it is not
"possible for it to remain always the same", and the referent
of the 'it' is "the now which appears to divide the past and the
future". Do those elementary observations about the need for
more than one limit for a period of time really weigh against
the view that the now which appears to divide the past and the
future is always one and the same? I believe they do, and very
heavily.
Aristotle may be taken to be saying something like this.
Every now which appears to divide past and future is an
instant, every instant of the past was such a now, and every
instant of the future will be such a now. The claim that the
now which appears to divide the past from the future is always
ARISTOTLE ON THE INSTANT OF CHANGE 10og
one and the same is thus tantamount to the claim that there is
only one instant, which is absurd.
Would it occur to anyone to make this claim about the now
in such a way as to leave himself open to so obvious a refuta-
tion? My guess is that someone who is convinced that time is
essentially passing and that the past and the future do not
exist, and who wants, all the same, to maintain that time
exists, might make such a claim. Time, he might say, is one
and the same in virtue of being "continuous", and its exist-
ence consists in the steady, continuous passage of the now
conceived of as the leading edge of becoming. Time exists as
the real passage of a single continuously moving now. People
certainly have spoken in that way,2 and it might be objected
that in this first argument, with its insistence on treating the
now simply as a temporal limit, Aristotle fails to address such
a view of the now. Perhaps he does, but then at least he re-
quires it to identify itself more precisely. The notion of the
now arises as the notion of the division between the past and
the future, a temporal limit. A temporal limit may be thought
of as an edge, but not as a leading edge; if it's a nunc, it's not
fluens. There are infinitely many temporal limits, but if there
is a leading edge in time, there is only one. And so it will not
do to say-if that is what this claim is out to say-that the
leading edge of temporal passage is a now.
Even if I have succeeded in showing how this first argument
makes contact with the view that the now which appears to
divide the past and the future is always one and the same, it
may be objected that it attacks it on a technicality, and a
technicality which it is particularly easy to stumble over in a
language which uses only one word for 'instant' and 'now'.
The second argument, I think, remedies this flaw by attacking
the view in a way that owes nothing to technicalities.
In addition, if to be simultaneous and neither earlier nor

later with respect to time is to be at one and the same
instant, then if both the earlier and the later were to be
at this now [we are discussing], what happened ten
thousand years ago would be simultaneous with what is
happening to-day, and nothing would be either earlier
or later than anything else (2 18a25-30).
The first argument charged this view with harbouring the

absurd notion of a unique temporal limit, but "this now [we
are discussing]" may not, perhaps, be fairly and completely
described as a temporal limit. This second argument, how-
ever, levels a charge which the view cannot evade-namely,
that it introduces an enduring now (appropriate only to a
timeless mode of existence, eternity). If what we are discussing
is always one and the same, and if it is truly now, and if
simultaneity is being at one and the same now, then every
event is simultaneous with every other event. Although
Aristotle introduces this absurdity in terms of temporal
dynamics-"ten thousand years ago" and "to-day"-he does
not stop with deducing an absurdity in temporal dynamics-
e.g., and so nothing would be either past or future-but
presses on to the fundamental absurdity in temporal statics,
the denial of the temporal order-"nothing would be either
earlier or later than anything else".
The unquestioned basis of both arguments in the develop-
ment of the fourth puzzle is temporal statics. For any two
events, either the one is earlier than the other or the two are
simultaneous; not all events are simultaneous; time is infinite
and continuous, but it may be divided into finite periods, the
limits of which are instants. These propositions of temporal
statics are not only relied upon and unchallenged in the dis-
cussion of the four puzzles, they also constitute the description
of the essence of time as viewed in Aristotle's own theory, as I
hope to show another time.m
APPENDIX (Physics IV, lo; 217b29-218a33)
b29 Next after what has been said comes the discussion
of time. Concerning time we would do well, making use
of the popular discussions also, to go over the puzzles
regarding its existence or nonexistence and then its
nature.
b32 That time is either altogether nonexistent, or that it
exists, but hardly, or obscurely, might be suspected from
the following considerations.
b33 One part of it has come to be but is not, the other
part will be but is not yet; and it is of these two parts
that infinite time, or any time one might take, is com-
ARISTOTLE ON THE INSTANT OF CHANGE I1
posed. But that which is composed of nonexistents might

be thought to be incapable of sharing in existence.
a3 In addition, if any thing with parts is to exist, then,
when it exists, all or some of its parts must exist. But,
although time is divisible, some parts of it have been
and the others will be, and no part of it exists. And as for
the now, it is no part of time; for a part measures, and
the whole must be composed of the parts; but it is not
thought that time is composed of instants.
a8 Again, it is not easy to see whether the now which
appears to divide the past and the future alwaysremains
one and the same or is always distinct.
al If it is always distinct, while no two parts of time
exist simultaneously unless one part contains while the
other is contained, as the smaller period of time is con-
tained by the greater, and if/
/(W-A) an instant which does not exist but existed
earlier must at some time have been
destroyed/
a4 that which does not exist now but existed
]/(C-R) earlier must at some time have been
destroyed/,
t en two instants cannot exist simultaneously, but always
a16 the earlier instant must have been destroyed. Now the
earlier instant cannot have been destroyed at that very
instant itself, because it existed then. And it cannot have
been destroyed at some later instant. For let us lay it
down that instants, like points on a line, cannot be
a19 consecutive to each other. Then if indeed it was not
destroyed at the succeeding instant but at another, it
would have existed simultaneously with the infinitely
many instants between itself and that other, and that is
impossible.
a2 1 Moreover, neither is it possible for it to remain always
the same; for no finite and divisible thing, whether
continuous in one or in more than one dimension, has
just one limit, but an instant is a limit, and it is possible
a25 to cut off a finite time. In addition, if to be simultaneous
and neither earlier nor later with respect to time is to
be at one and the same instant, then if both the earlier
112 II----NORMAN KRETZMANN
and the later were to be at this now [we are discussing],

what happened ten thousand years ago would be simul-
taneous with what has happened to-day, and nothing
would be earlier or later than anything else.
a3o Let this, then, be the discussion of puzzles faced in
connexion with what belongs to time.
But what time is and what its nature is is as unclear
from the accounts handed down to us as it is from the
puzzles just discussed.
NOTES
1 Both Hesychius' and Ptolemy's ancient lists of Aristotle's works contain

references to a treatise on time, which Ptolemy describes as consisting of one
book. Ross believes that those references "probably refer to the essay on time
in Phys. iv. 10-14, which may well have been originally a separate treatise"
(Aristotle's Physics, p. 6). And the discussion of time begins and ends with
formulae of the sort Aristotle uses to open and close treatises. (A translation
of the puzzles passage appears as an appendix to this article.)
2 Ross discusses this phrase in his commentary
(pp. 595-596) and reviews
interpretations of Aristotle's uses of the phrase in his commentary on
Metaphysics lo76a28 (pp. 408-410). The logoi might be either claims or
arguments, and the designation 'exoterikoi' might mean either of the sort
encountered outside the Lyceum or published in one of Aristotle's (lost)
dialogues. Ross says of this particular occurrence of the phrase that it seems
that "discussions rather than any special books are meant-discussions not
peculiar to the Peripatetic school; but in many cases Aristotle had in point
of fact developed these in his dialogues. The logoi here referred to are
those that are put forward in b33-218a3o" (A's Phys., pp. 595-596). If the
logoi are indeed the arguments used to work up the four puzzles, I think
they must at least have been "developed" by Aristotle. Since the puzzles con-
stitute difficulties for the view of time held by Plato (among many others), the
arguments which support them are very unlikely to have come out of the
Academy, and there seems to be no other plausible "exoteric" source for argu-
ments with the aim and level of sophistication these have. Because of these
considerations and because the view of time for which the puzzles constitute
difficulties is not only Platonic but also intuitive, I am inclined to think that
the exoteric logoi are familiar claims rather than arguments with their source
outside the Lyceum.
3 See,
e.g., G. E. L. Owen in his "Aristotle on Time" (see n. 27 inf.):
"Aristotle evidently thought his paradoxes could be evaded . . ." (p. 8);
"There is a question here: given that the paradoxes are not systematically
and directly answered in the sequel, how and when were they prefaced to the
argument? . . ." (n. 27). As I shall try to show, however, the puzzles
do not cut against Aristotle's own position but rather prepare the ground for
it. The fact that he proceeds without considering them further indicates only
that he takes them to have served their purpose. There is no more reason to
look for Aristotle's answers to these puzzles than there is to look for Zeno's
solutions to his paradoxes.
ARISTOTLE ON THE INSTANT OF CHANGE 1 13
4 Richard Sorabji has suggested to me that the phrase 'ho aei lambanomenos
chronos' might be read as 'the time that is ever being taken' (rather than as
'any time one might take'). On that reading the phrase appears to refer to
refer to any stretch of "present time"--i.e., to any N-period.
5 For purposes of this discussion I am not questioning the legitimacy of
describing the past simply as that which is no longer and the future simply
as that which is not yet, or of describing the past and the future as therefore
nonexistent or unreal. I am grateful to Carl Ginet for a helpful suggestion
regarding this first objection.
6 "Aristotle on the
Reality of Time", Archiv fiur Geschichte der Philosophie
56 (1974)) 132-155; p. 132. Miller's article is the best discussion I have seen
of the topics considered in this article. Although our interpretations differ in
many respects, I have learned a great deal from him.
?The description here is actually only "has come to be and is not" (gegone
kai ouk estin / 217b34), but what has come to be and is not is no longer.
8 The most famous example of this stripping away is Augustine's in
Confessions Book XI, chs. 19-2o, the passage which Miller has christened
"the whittling argument" (op. cit., p. 133).
- Greek uses one word -'nun'-adverbially in the sense of 'now' and sub-
stantivally in the sense of 'instant' or, more particularly, 'the present instant',
'the now'. In the puzzles passage it is frequently important and difficult to
decide on the correct interpretation, which is rarely made perfectly
unambiguous by the context.
10The force of the "it is . . . thought that" (dokei) cannot be to stigmatize
this premiss as mere uninformed opinion; it presents an essential feature
of Aristotle's own view of the temporal continuum. Perhaps it can be read
instead as an indication that the premiss is beyond serious dispute.
11I am grateful to Richard Boyd for a helpful suggestion regarding these
necessary characteristics.
12Miller, op. cit., p. 150; quoting Gale in the latter half of the passage.
Such imagery goes back to Broad and McTaggart at least.
13Such descriptions are misleadingly specific because they involve spatially
separated sentient beings. They may thus lead us to think of what are sup-
posed to be merely distinct parts of time as distinct events, and there is no
relevant difficulty in conceiving of two distinct but completely coincident
events.
14What if we choose as the second period the overlapping three-hour
period from ten until one to-day? Does Aristotle's description omit that sort
of case from consideration? No; the only relevant second period in this case
is the two-hour period from ten until noon to-day. The question of distinct
and simultaneous parts of time does not arise in connexion with the third
hour of the overlapping period, or with the three-hour overlapping period
considered only as a whole.
15Taking time to be linear involves taking it to be one-tracked. In this
connexion see 218b5, where Aristotle reduces a predecessor's opinion to an
absurdity by showing that it allows the possibility that "a plurality of times
would exist simultaneously".
16 If it had suited his
purposes, Aristotle could have taken a very short
way with this thesis. As he points out in the development of puzzle IIIb
(218a25ff.), "to be simultaneous . . . is to be at one and the same instant";
and so a "third time" argument could have been generated here. Cf. 218b17:
"time is not defined in terms of time".
17 P. H. Wicksteed
(Loeb edn., g929): "the 'now' that is not, but was, must
have ceased to be at some time or other"; H. G. Apostle (Indiana U.P., 1969):
"a moment which does not exist but existed before must have been destroyed
sometime".
1a Cornford's reading appears in his editorial footnote to Wicksteed's
translation: "anything that does not exist now, but formerly did exist, must
have perished at some moment"; Ross's interpretation appears in his Analysis
of this passage: "that which now is not and previously was must have ceased
to be" (A's Phys., p. 384).
19As Wicksteed points out in a note to the text here, "This is proved in
Book VI, chap. i" (231b6-18).
a0See my "Incipit/Desinit" (see n. 28 inf.). Strictly speaking, instantaneous
existence is a special case of "permanent" existence; it is therefore not always
separately recognized by the logicians who discuss such topics.
2 See
p. 98 above.
22 Instants in the temporal continuum, like points on a line, are successive,
distinct, and nonconsecutive. But our concern here is not with instants
generally but with "the now which appears to divide the past and the
future", and if that now is to be analysed into successive distinct nows, each
of which begins to exist, exists, and ceases to exist instantaneously, then there
must be one such now after another.
a Temporal dynamics and temporal statics are associated, respectively, with
the relationships McTaggart designated "the A series" and "the B series";
I prefer to call them 'PNF' (for past-now-future) and 'ESL' (for earlier-
simultaneous-later). All events are permanently ordered in terms of ESL; all
and only events within an N-period are transitorily ordered in terms of PNF.
24 I hope to try to substantiate these claims in a later paper. They seem to
deserve at least a close look. As Miller points out, "It is a commonplace that
Aristotle treats time as something inherently dynamic" (op. cit., p. 147); and
in a note to that sentence he says, "For a contrasting view cf. W. Wieland
[Die aristotelische Physik (G6ttingen 1962)], p. 327: "Vor allem versteht
Aristoteles die Zeit nicht von einem 'flieszenden' Jetzpunkt aus, wie es ihm
immer wieder unterstellt wird".
2 Plato is one Aristotle
may have heard speak in that way; Donald Williams
provides a colourful sampler of more recent remarks of the sort in "The Myth
of Passage", The Journal of Philosophy 48(1951).
26 I want to supplement the specific acknowledgements made in earlier notes
by expressing special thanks to Fred Miller, who read and carefully criticized
the seminar notes from which this paper evolved, to Gail Fine, Terence
Irwin, Malcolm Schofield, and Richard Sorabji, for their thoughtful criticisms
of an earlier draft, and above all to Eleonore Stump, whose help was
unstinting and invaluable.
27 In Machamer & Turnbull edd., Motion and Time,
Space and Matter
(Columbus: Ohio State U.P., 1976), pp. 3-27.
28
In Machamer & Turnbull, op. cit., pp. 101-136.

Sorabji-Kretzmann, Aristotle On The Instant of Change

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sorabji-Kretzmann, Aristotle On The Instant of Change

Uploaded by

Copyright:

Available Formats

Aristotle on the Instant of Change

Author(s): Richard Sorabji and Norman Kretzmann

thing as a time-atom, that is, an indivisible period with an

one version of the problem, and if this version were inapplic-

Proposed solution for casesof continuous change

of transition between rest and continuous motion as an instant

Further support for this decision is gained, if we consider a

always in its turn change continuously, or can it jump dis-

The solution will apply not only to continuous motion, but

Treatment of other cases

existence to existence, or from invisibility to visibility. In

Aristotle's treatment. Preliminaries (i): The four main kinds

Preliminaries (ii): How can change of colour be gradual?

putting things seems to neglect the idea that there will be a

Aristotle's first solution

Aristotle's second solution

one of the passages where he does not construe the process as

tioned,whereAristotleis attractedto a view close to our own.

Assessment of Aristotle's first solution

behaviour. Much of modern dynamicsdepends on the possi-

1 This revival has been discussed by Curtis Wilson in William Heytesbury,

(Physics IV, 10o;217b29-218a33)

The Four Puzzles

IIIb. there are differentsuccessivenows-which is

now (but not as either of the limits of the time); suppose we

even if he agrees that the present must be conceived of as a

(i) A part must measure the whole

The differencesseparating(i) and (ii) from (1) and (2) are

of the distance covered as of now or the time elapsed as of

conclusion may seem to be a truism; of course there is no time

The choice is an odd one, perhaps odder than it looks at

be led to think casuallyin terms of simultaneous distinct nows.

nine until noon to-day is a part distinct from but simultaneous

his discussion of time, to say expressly that the sort of simul-

tantalizingly close to the actual text, but I am not bold enough

taneously with the infinitely many instants between itself

used a bad or at least a suspect argument where a good one lay

seem designed to show that one cannot consistently hold that

In addition, if to be simultaneous and neither earlier nor

The first argument charged this view with harbouring the

posed. But that which is composed of nonexistents might

and the later were to be at this now [we are discussing],

1 Both Hesychius' and Ptolemy's ancient lists of Aristotle's works contain

You might also like