Further, everything that changes must be divisible. For since
every change is from something to something, and when a thing is at
the goal of its change it is no longer changing, and when both it
itself and all its parts are at the starting-point of its change it is
not changing (for that which is in whole and in part in an unvarying
condition is not in a state of change); it follows, therefore, that
part of that which is changing must be at the starting-point and
part at the goal: for as a whole it cannot be in both or in neither.
(Here by "goal of change" I mean that which comes first in the process
of change: e.g. in a process of change from white the goal in question
will be grey, not black: for it is not necessary that that that
which is changing should be at either of the extremes.) It is evident,
therefore, that everything that changes must be divisible.

    Now motion is divisible in two senses. In the first place it is
divisible in virtue of the time that it occupies. In the second
place it is divisible according to the motions of the several parts of
that which is in motion: e.g. if the whole AG is in motion, there will
be a motion of AB and a motion of BG. That being so, let DE be the
motion of the part AB and EZ the motion of the part BG. Then the whole
DZ must be the motion of AG: for DZ must constitute the motion of AG
inasmuch as DE and EZ severally constitute the motions of each of
its parts. But the motion of a thing can never be constituted by the
motion of something else: consequently the whole motion is the
motion of the whole magnitude.

    Again, since every motion is a motion of something, and the whole
motion DZ is not the motion of either of the parts (for each of the
parts DE, EZ is the motion of one of the parts AB, BG) or of
anything else (for, the whole motion being the motion of a whole,
the parts of the motion are the motions of the parts of that whole:
and the parts of DZ are the motions of AB, BG and of nothing else:
for, as we saw, a motion that is one cannot be the motion of more
things than one): since this is so, the whole motion will be the
motion of the magnitude ABG.

    Again, if there is a motion of the whole other than DZ, say the
the of each of the arts may be subtracted from it: and these motions
will be equal to DE, EZ respectively: for the motion of that which
is one must be one. So if the whole motion OI may be divided into
the motions of the parts, OI will be equal to DZ: if on the other hand
there is any remainder, say KI, this will be a motion of nothing:
for it can be the motion neither of the whole nor of the parts (as the
motion of that which is one must be one) nor of anything else: for a
motion that is continuous must be the motion of things that are
continuous. And the same result follows if the division of OI
reveals a surplus on the side of the motions of the parts.
Consequently, if this is impossible, the whole motion must be the same
as and equal to DZ.

    This then is what is meant by the division of motion according to
the motions of the parts: and it must be applicable to everything that
is divisible into parts.

    Motion is also susceptible of another kind of division, that
according to time. For since all motion is in time and all time is
divisible, and in less time the motion is less, it follows that
every motion must be divisible according to time. And since everything
that is in motion is in motion in a certain sphere and for a certain
time and has a motion belonging to it, it follows that the time, the
motion, the being-in-motion, the thing that is in motion, and the
sphere of the motion must all be susceptible of the same divisions
(though spheres of motion are not all divisible in a like manner: thus
quantity is essentially, quality accidentally divisible). For
suppose that A is the time occupied by the motion B. Then if all the
time has been occupied by the whole motion, it will take less of the
motion to occupy half the time, less again to occupy a further
subdivision of the time, and so on to infinity. Again, the time will
be divisible similarly to the motion: for if the whole motion occupies
all the time half the motion will occupy half the time, and less of
the motion again will occupy less of the time.

    In the same way the being-in-motion will also be divisible. For
let G be the whole being-in-motion. Then the being-in-motion that
corresponds to half the motion will be less than the whole
being-in-motion, that which corresponds to a quarter of the motion
will be less again, and so on to infinity. Moreover by setting out
successively the being-in-motion corresponding to each of the two
motions DG (say) and GE, we may argue that the whole being-in-motion
will correspond to the whole motion (for if it were some other
being-in-motion that corresponded to the whole motion, there would
be more than one being-in motion corresponding to the same motion),
the argument being the same as that whereby we showed that the
motion of a thing is divisible into the motions of the parts of the
thing: for if we take separately the being-in motion corresponding
to each of the two motions, we shall see that the whole being-in
motion is continuous.

    The same reasoning will show the divisibility of the length, and
in fact of everything that forms a sphere of change (though some of
these are only accidentally divisible because that which changes is
so): for the division of one term will involve the division of all.
So, too, in the matter of their being finite or infinite, they will
all alike be either the one or the other. And we now see that in
most cases the fact that all the terms are divisible or infinite is
a direct consequence of the fact that the thing that changes is
divisible or infinite: for the attributes "divisible" and "infinite"
belong in the first instance to the thing that changes. That
divisibility does so we have already shown: that infinity does so will
be made clear in what follows?