Conservation

as related to
Continuity and Constancy

1 Conservation : The Basic Idea

When electric charge flows from place to place, it obeys a local conservation
law. By that we mean something very specific:
Any decrease in the amount of charge in a given region of space
must be exactly balanced by
a simultaneous increase in the amount of charge in an adjacent region of
space.

We can express the same idea in the form of an equation:

change in
= net flow of charge
charge
(1)
(inside (inward minus outward across
boundary) boundary)

The word “flow” in this expression has a very precise technical meaning,
closely corresponding to one of the meanings it has in everyday life.
Specifically: Charge cannot be created or destroyed, but it can flow one from
one region to another. Any decrease in one region is exactly balanced by a
simultaneous increase in an adjacent region.

(By way of contrast, if you want an example of something that flows but
is not conserved, consider “the flow of ideas”. You can give me an idea without
decreasing your own stock of ideas. As usual, flow refers to spreading from
one region to a neighboring region.)
At this point we have three ideas on the table: conservation, change, and flow.
In accordance with equation 1, we can understand conservation as a
combination of change plus just the right amount of flow.

To repeat: electric charge cannot change arbitrarily; only certain types of

changes are allowed. The relationship between the broad concept of “change”
and the more-precise concept of “conservation” can be summarized as follows:

Conservati chan equal-and-opposite change +

= +
on ge simultaneity + adjacency (
chan 2)
= +matching flow
ge

The use of the world “simultaneity” causes some people to wonder

whether equation 2 can be relativistically correct. Well, it turns out that this
equation is 100% kosher, relativistically and otherwise. We get into trouble
with relativity when we talk about simultaneity at a distance, but since we are
taking about adjacent changes – zero distance – there is no problem with
simultaneity. Indeed, our notion of conservation is beautifully matched to
modern notions of relativity. In spacetime, local conservation can be
expressed in elegant, informative, and easy-to-visualize form, in terms of
the continuity of world lines. This is discussed in section 2.
To be useful – or even correct – the conservation law must be local, as
discussed in section 2.3. It must apply to every region in spacetime, including
small open regions where the charge is not constant. Charge
is always conserved, even though it is not always constant, as discussed
in section 2.4.

As a corollary, as a special case of equation 2:

• If there is no flow across the boundary, conservation implies there

cannot be any change.

• If there is no flow across the boundary, conservation implies constancy.

This stands in contrast to the general case, where conservation does not imply
constancy, nor vice versa. This is discussed in section 2.4.
To summarize: The laws of physics do not permit charge to disappear now and
reappear later. Similarly the laws do not permit charge to disappear from here
and reappear at some distant place.

Charge is conserved right here,

right now.

2 Conservation in Spacetime

2.1 Conservation
An ordinary box in D=3 space has six faces. We can call them the ± x faces, the
± y faces, and the ± z faces. In D=4 spacetime, the corresponding object has
eight faces, namely two ± t faces in addition to the six spatial faces.

Now let’s draw some spacetime diagrams. For simplicity we will conceal
the y and z dimensions, and portray only the ± t and ± x faces of the box.

Figure 1 shows the world-line of a stationary parcel of electrical charge sitting

at the location of our box. The world-line enters on the t=0 face and exits on
the t=1 face. In other words, it was initially within the spatial extent of our
box, and it remained so.
Figure 1: World-line of Stationary Parcel

World-lines crossing the ± t faces contribute to the time derivative of the

amount of charge in the box. In figure 1, there are two equal and opposite
contributions, namely one flowing in at t=0, and one flowing out at t=1.

In 4-dimensional spacetime In 3-dimensional spatial terms:

terms: The charge is flowing in The charge is not flowing at all.
the time direction only. It just sits there. The amount of
charge within the spatial extent
of the box is the same at t=0
and t=1, so the amount of charge
in that spatial region is constant,
i.e. not changing with time.

We now consider what happens when the charge is non-stationary, as shown

in figure 2. The world-line enters via the t=0 face and exits via the x=1 face. In
other words, the parcel was initially within the spatial extent of our box, but
then it departed toward the right.
Figure 2: World-line of Moving Parcel

In this case, the amount of charge in the spatial box at t=1 is less than the
amount of charge at t=0. The charge is not locally constant, but it is locally
conserved. The charge wasn’t destroyed, it just flowed across the x=1
boundary into the neighboring region.
We will make this more quantitative in a moment. To make this useful, we will
need to quantify the net flow across the ± x boundaries. It helps to know
something about velocity and charge density.

You can see where this line of reasoning is leading. The following notions are
formally equivalent:

• Local conservation.

• Continuity world-lines.

2.2 Non-Conservation

We now consider a non-conserved object, such as a muon. Here is the world-

line of a muon that decays within the box; the world line enters the box but
doesn’t go out again:
Figure 3: World-line of Non-Conserved Object

Here are some trickier examples of non-conservation. In figure 4, the particle

disappears and then reappears at a later time. Electric charge does not behave
like this. The law of conservation of charge forbids it. In particular, if we
divide the region into an early half and a late half, we now have two regions,
both of which exhibit gross non-conservation. The lesson here is that the
conservation is not equivalent to saying that you end up with the same
amount you started with. The conservation law applies not just at the starting
time and ending time, but at every time along the way.

Figure 4: Non- Figure 5: Non-

Conservation: t Gap Conservation: x Gap

Similarly, in figure 5, the particle disappears at one location and reappears at

some other location. Electric charge does not behave like this. The law of
conservation of charge forbids it. In particular, if we divide the region into left
half and a right half, we now have two regions, both of which exhibit gross
non-conservation. The lesson here is that the conservation is not equivalent to
saying that the total amount is constant over time. The conservation law
applies not just to the universe as a whole, but also to every region and
subregion.
2.3 Locality

It must be emphasized that conservation of charge is a local conservation law.

The property of “locality” is crucial, for at least two reasons, as we shall see.

It is possible to have global conservation without local conservation, as shown

in figure 5. If we consider only the large region from x=0 to x=1, there is no
change in the charge inside, and no flow across the ±x boundaries, so we have
conservation in accordance with equation 1.

The first huge problem with global conservation is that it is not useful in
practice. If some charge disappears from your laboratory and reappears
somewhere far, far away, you have no way of knowing whether it is really
conserved. You have no way of checking. An unenforceable law is useless.

The second huge problem manifests itself when we consider a second

reference frame, moving relative to the first, in accordance with the laws of
special relativity. The situation is shown in figure 7. In this figure, the world
lines are exactly the same as in figure 6 (and also figure 5); only the reference
frame is changed. Whereas an ordinary rotation in the xy plane rotates
the x and y contour lines the same way, a spacetime rotation in the xt plane
tilts the x contours one way and the t contours the other way. For details,
see reference 1. In figure 7, the amount of tilt corresponds to a x^t^ frame that
is moving at half the speed of light (relative to the xt frame).

Figure 6: Non- Figure 7: Non-Conservation: x Gap,

Conservation: x Gap Boosted

The point of all this is that in the x^t^ frame, the particle disappears later than
time t^=0.5 and reappears earlier than that. Therefore even though the
amount of charge is globally constant relative to the xt frame, it is not
constant relative to the x^t^ frame. Since the laws of physics are required to
work in any reference frame, we see that any notion of non-local conservation
is dead on arrival. In the real world, i.e. in spacetime, non-local conservation is
almost no conservation at all.1

2.4 Conservation versus Constancy

It is important to distinguish the notion of constancy from the notion

of conservation:

• When we say some quantity q is constant, or equivalently unchanging, it

means simply that the amount of q in the region does not change. There
are various ways this could happen, either with or without
conservation, as spelled out in the table below.

• When we say some quantity q is conserved, it means that the amount

of q does not change except by flowing across the boundary of the
region. Conservation can happen with or without change, as spelled out
in the table below.

Imagine a fluid that has the same velocity and the same charge-density
everywhere. An example of such a situation is shown in figure 8.

Figure 8: Constancy via Replacement of Conserved Particles

In this case, the amount of charge in the box is constant, because whenever a
parcel of charge leaves via the x=0 face, an equivalent amount of charge is
entering via the x=1 face. Contrast this with figure 2, where the charge flowed
out and was not replaced. This non-replacement could occur because the
charge density was lower off to the left, and/or the velocity was lower off to the
left. So we see we need to be concerned with

• the charge density

• the velocity

• the left-to-right variation thereof.

A particularly interesting case of constancy with flow and replacement is

steady flow around a loop, as shown in figure 9. The diagram is subject to
periodic boundary conditions. That is, the two dashed worldlines are
the same worldline. Also the two dashed red boundaries are
the same boundary. Every time a particle flows out of one compartment it is
replaced by another particle flowing in.

Figure 9: Constancy and Conservation: Steady Flow Around a Loop

We have just seen constancy with conservation, via outflow compensated by

inflow. We now constrast that against constancy without conservation, via
annihilation and inflow. This shown in figure 10. You can see that the number
of objects in the region of interest is constant, because every time one of them
is destroyed it is replaced.

Figure 10: Constancy via Replacement of Non-Conserved Particles

A real-life example of this concerns the number of working light-bulbs in a

room. The number of working light bulbs is not a conserved quantity, but the
number is approximately constant, because every time a light bulb is
destroyed you replace it.
Another example along the same lines is a continuous-flow reactor in a
chemical plant, where chemicals of type A flow in and chemicals of type B flow
out.

To be explicit: Conservation does not imply constancy, nor vice versa. Change
does not imply non-conservation, nor vice versa. All four of the possibilities
have already been illustrated:

constant non-constant

conservation Figure 1 (static) Figure 2 (outflow)

Figure 8 (steady flow)

non-conservation Figure 10 (steady replacement) Figure 3 (annihilation)

As touched upon section 1, if you integrate the local conservation law over all
space, you can derive a global conservation law as a corollary. It is true that
global conservation is the same as global constancy ... but that does not mean
that conservation is the came as constancy in general.

Constancy can be used as a proxy for conservation for closed systems only. By
far the more powerful idea is conservation (and local conservation in
particular), because it can be applied to small, open sytems. In particular,
suppose we have a small region. Charge is flowing across the boundary.
Charge is conserved, even though the charge in our region is not constant. We
could try to salvage the idea of constancy by considering a larger region, but
then we lose the idea of locality ... and without locality we are doomed, for
multiple reasons, as discussed in section 2.3.
In summary: The only way to do the whole job properly is to define a notion of
conservation that works for small, open systems – that is, local conservation,
namely equation 1. The key ideas are:

• To be useful, the law must be local. Furthermore, special relativity says

that a non-local law cannot possibly be correct.

• Locality requires us to consider small systems.

• Small systems are generally not closed.

• Therefore a law that applies only to closed systems is nowhere near

strong enough to be a fundamental law of physics.

• The fact is, charge is always conserved ... even in small, open regions
where the charge is not constant. This is true and important.

3 Discussion

3.1 Many Are Conserved

We have been using charge as our primary example of a conserved quantity.

There is a long list of such things.
 • Charge is conserved.

• The [energy, momentum] 4-vector is conserved.

• If we decompose the [energy, momentum] 4-vector into components in
any chosen basis, the x, y, and z components of momentum are
separately conserved. This can be used to derive a useful equation of
motion for fluids, namely the Euler equation; see reference 2.

• Similarly, in any chosen basis, the energy is separately conserved. This

makes sense in relation to the previous item, given that the energy is
the fourth component of the [energy, momentum] 4-vector. Conservation
of energy is synonymous with the first law of thermodynamics, as
discussed in reference 3.

• Lepton number is conserved.

• Et cetera.

The picture of world-line continuity applies to them all.

In addition, there is an even longer list of things that are approximately

conserved.
• Entropy obeys an equation very similar to equation 1, but it is an
inequality rather than an equality.

• If we restrict attention to reversible processes, entropy obeys a

conservation law. (This is not an exact law of nature, insofar as nothing
is exactly reversible.

• The mass of a particle is not conserved.

Mass is a relativistic invariant, i.e. a Lorentz scalar, related to the invariant

norm of the [energy, momentum] 4-vector. For a particle with nonzero mass,
this is proportional to the rest energy in accordance with the celebrated
formula Erest = m c2. For a slow-moving massive particle, the energy is very
nearly equal to the rest energy, which implies that the mass is conserved to a
good approximation (but not exactly).

• In a chemical reaction, the number of atoms of a given type is

conserved. (This is what distinguishes chemical reactions from other
things such as nuclear reactions.) If there are 100 different kinds of
atoms involved, this gives us 100 separate conservation laws, one for
each atom ... in addition to all the non-chemical conservation laws
(energy, charge, et cetera).

3.2 Warning: Conflicting Terminology

Beware: The term “conservation” has two wildly different meanings.

The physics notion of The homespun notion of

conservation is the topic of this conservation refers to protecting
document. It is defined by
 expressions of the something against destruction,
form equation 1. or especially against waste.

Example: Charge is always Example: Conservation of

conserved – automatically, endangered wildlife. People get
strictly, and locally – in to decide whether they want to
accordance with equation 1. conserve this-or-that. Sometimes
they decide yes, sometimes no.

While we’re on the subject: Beware: There are two different notions of energy.
There exists a homespun notion of «useful energy» that is important and
widely used, yet is very different from the physics energy.

The physics energy includes all The homespun energy refers to

forms of energy, whether useful some ill-defined notion of “useful
or not. energy” or “available energy”.

Example: The rest energy is Examples: The energy industry.

given by E=mc2. The Department of Energy.

Suppose you see a statement of the form “Please conserve energy by turning
out the lights”. Be careful not to intepret that in physics terms, because both
of the key words – “conserve” and “energy” – are being given meanings wildly
different from their physics meanings.

3.3 The Current is Not Conserved

Just because electrical charge is conserved does not mean that electrical
current is conserved. In fact, the current is definitely not conserved. Imaging a
loop of wire with some current flowing in it initially. If you just leave it alone,
the current will die out. The current will not move into some neighboring
region; it will just remain in place and get weaker. This is an absolutely clear
example of non-conservation.
This is mostly just a terminology issue, because physically the current is
intimately associated with the thing that is conserved. Indeed, you cannot
even write down the equation for conservation of charge without mentioning
the current also:

∂
ρ + ∇·
0 = (3)
j
∂t

We say that the current obeys an approximate continuity equation (not a

conservation equation, approximate or otherwise). More specifically, in the
low-frequency limit – for so-called “DC” circuits – we can ignore the first term
on the RHS of equation 3, and what’s left is a continuity equation for the
current, namely equation 4:

∇· (3D continuity of DC
0 ≈ (4)
j current)
 (implies no
accumulation)

One of Kirchhoff’s «laws» says that current obeys a continuity equation.

People very often design circuits so that they uphold Kirchoff’s laws, to a good
approximation ... but this is a choice, not a law of nature, as discussed
in reference 4.

While we’re in the neighborhood, if we instead ignore the other term on the
RHS of equation 3, we get:

∂
ρ (constancy of
0 ≈ (5)
charge)
∂t

Zero accumulation does not imply zero flow. It just means no net flow across
the boundaries. You can have lots of flow-with-replacement. You can also have
lots of flow going round and round in the interior, not crossing the
boundaries.

In fluid dynamics, vortex lines obey a continuity equation. In electrodynamics,

magnetic field lines obey an exact continuity equation in three dimensions. In
any region where there is no electrical charge, electric field lines obey a
continuity equation.

The important idea is that the full conservation equation – equation 3 – can be
understood as a spacetime continuity equation, as discussed in section 2.

In some weird sense, the idea of constancy (equation 5) is an amputated part

of the idea of conservation, and also the idea of three-dimensional continuity
(equation 4) is an amputated part of conservation. These two terms working
together give the proper idea of conservation, i.e. four-dimensional continuity
(equation 3).

Bottom line: You shouldn’t talk about “conservation” of the current vector;
that gives entirely the wrong idea. If you want to say something about the
current, talk about the spacetime continuity of the [charge, current] 4-vector.

4 Quantifying Conservation

Now that we have a firm qualitative idea of conservation, let us now make the
idea completely quantitative.

4.1 Differential Version

For an electrically-charged fluid, the product of charge density times velocity

is called the current. A spatially-uniform current isn’t interesting from a
conservation point of view, so we shall be particular interested in the spatial
variation in the current. If there is more than one spatial direction, the
current is a vector (just like the velocity) with x, y, and z components. What
matters is the x-derivative of the x-component of flux, the y-derivative of the y-
component, and the z-derivative of the z-component — in other words, the
divergence.

So we can write the local conservation law as

∂
(charge density) = − div(current density) (6)
∂t

or equivalently

∂
ρ + ∇·j = 0 (7)
∂t

This is a 100% formal accurate statement of the local conservation law. It

expresses the continuity of the world-line of each little bit of charge.

Homework: Check equation 7 using dimensional analysis. It shouldn’t take

very long!

The structure of equation 7 is very elegant. It has the form

∂ ∂ ∂ ∂
(...) (...) (...) (... ..
= (8)
∂ + ∂ + ∂ + ∂ ) .
t x y z

which is most elegantly understood as a four-dimensional divergence. If you

leave off the ∂/∂t term you’ve got the plain old three-dimensional divergence.
The key concept is that the world lines of a conserved quantity (such as
charge) are divergence-free in D=4. They have no beginning or end. By way of
contrast, if you projected the world lines onto three dimensions, they would
not be divergence-free; there would be a nonzero divergence at any location
where charge is accumulating. The timelike term in equation 8 is needed to
make the equation correct. On the third hand, we can contrast this with
magnetic field lines, which are divergence-free – i.e. endless – even in three
dimensions.

In more detail: The world line of an element of a charge has a tangent vector
at every point. This defines a vector field. At each point along the world line,
there is a four-vector consisting of the three components of current plus a
fourth component, representing the “flux” of charge flowing in the “time
direction” – which is just the density of charge itself. Overall, this vector field
is divergence-free in D=4.

Equation 7 is frame-independent. Each of the terms on the LHS is frame-

dependent, but the LHS as a whole is the same in all frames, the same for all
observers. The RHS is zero, which is manifestly invariant.
To repeat: In D=3 you can (temporarily) have a whole bunch of current
diverging from some region of space. [In contrast, you cannot have magnetic
flux lines diverging like this, not even temporarily.] The divergent current will
deplete the charge density in that region. Meanwhile, if you take a D=4 view of
the same situation, you will find that the [charge, current] four-vector has zero
divergence. The world-lines of any conserved quantity are endless.

In the preceding paragraph, all the D=4 statements are frame-independent, i.e.
Lorentz invariant. (As always, any D=3 statements are necessarily not Lorentz
invariant.)

The formulation in this section applies at a single point, or perhaps in the

infinitesimal neighborhood of a point. (A neighborhood is necessary to make
derivatives meaningful.) It involved density and flux, in particular the four-
dimensional divergence of the [charge-density, current-density] 4-vector.

Similar laws apply to the [energy, momentum] 4-vector, but the details are
messier. Note the contrast:

• For charge, the conserved quantity is charge itself. It is a scalar. The

current is a vector that is “associated” with the charge, but the current
is not per se the conserved quantity.

• For energy, things are quite a bit more involved.

Let’s choose a particular basis, and project out the components of the [energy,
momentum] 4-vector. We start by focusing attention on the energy component.
This is a conserved quantity, in much the same way that charge is a conserved
quantity. There is a slight difference, because charge is a Lorentz invariant
whereas energy is not, but let’s pretend we didn’t notice that. In any particular
chosen basis, that doesn’t matter very much.
Following the pattern of equation 8, we need four terms to express
conservation of energy: One for the time-dependence of the energy, and three
more for the divergence of the flow of energy.

Similarly, we need four terms to express conservation of the x-component of

momentum, and four more for the y-component of momentum, and four more
for the z-component of momentum.

Altogether, that means we need sixteen terms to express conservation of the

[energy, momentum] 4-vector. These can be arranged in a 4×4 array, i.e. a
second-rank tensor, namely the stress-energy tensor.

It takes a bit of work to keep track of how momentum is conserved, but it is

doable. The stress includes pressure, tension, and shear.

o Pressure involves things like x-momentum flowing in the

+x direction.

o Tension involves things like x-momentum flowing in the

−x direction.

o Shear involves things like the x-momentum flowing in the y-

direction, and z-momentum flowing in the x-direction, et cetera.
For more information about this, including suggestions on how to diagram the
flow of momentum, see reference 5.

4.2 Integral Version

Rather than considering a single point, we now consider a region.

We take the differential version (as discussed in the previous section) and
integrate it over a region. Using Stokes’s law, we find that the time-rate-of-
change of current inside the boundary is equal to the amount of current
flowing across the boundary.

The integral formulation implies the differential formulation and vice versa,
subject to mild restrictions. Obviously things must be sufficiently
differentiable, and the region of interest must be simply connected (no
wormholes).

5 Conservation and Continuity Without Constant Density

Consider the flow pattern shown in the following video. Charged particles
cross the boundary into our field of view at the left, flow from left to right, and
then exit our field of view. At no point are any particles created or destroyed.
There is strict conservation of charge and continuity of current.
Furthermore, this is a steady flow – not just conserved, but also steady. Steady
flow means the amount of charge crossing any given contour of constant x is
the same, independent of time, for all times between the beginning and end of
the movie. (If you imagine that the end of the movie portrays the charged
particles coming to a stop, that is an example of unsteady flow. Charge is still
conserved, but the flow is unsteady.)

Thirdly, you can see that the current is independent of x. This can be
understood as a corollary of the fact that the flow is both conservative and
steady.

Time 0

However, this does not mean that charge behaves like an incompressible fluid.
We have conservation of charge, continuity of current, and steady flow, but all
of that together is not sufficient to guarantee that the density of charge
remains constant.

Let’s be clear: constant density would be one way of guaranteeing that what
goes in must come out ... but it is by no means the only way, as you can see in
the video. Conservation of charge (and continuity of current) just means that
if the density decreases, the speed must increase, so that the current remains
the same.

Situations like what you see in the video happen all the time in the real world.
For example,
 • This happens whenever you make a junction between two different
types of metal; the carrier density is likely to be substantially different
in the two metals.

• The same is true to an even more extreme degree at the junction

between a metal and a semiconductor.

• You see the same sort of thing on the highway. Cars are conserved, in
the short term, to a decent approximation. When you come to a zone
with a high speed limit and good driving conditions, the cars spread out
and speed up. When they get to a zone with a lower speed limit and/or
worse conditions, they slow down and get closer together.

Again: Please do not imagine that charge behaves like an incompressible fluid.
A lot of people who ought to know better get this wrong. The fact is, in the
video, the density in the middle is a factor of two less than the density near the
ends.

Here’s another important distinction: If we watch at any fixed location along

the x axis, the density at that location is independent of time. In contrast, if we
follow a particular parcel of fluid as it flows along, the density of that parcel
particle changes dramatically. So if you ask whether this is a “constant
density” situation, the answer very much depends on which viewpoint you
take.

In particular, if you want Kirchhoff’s laws to be reliable, you need to restrict

attention to the DC limit, i.e. to steady situations. (In other situations, you can
engineer things so that the violations of Kirchhoff’s laws are small, but this is
an engineering choice, not a law of physics.) For details on this,
see reference 4. In such a situation, we know the charge density at any
particular point is constant because the flow is steady ... certainly not because
it is allegedly incompressible. It’s actually quite compressible.

When you see look at some chosen window of observation, constant density
does not tell you much about conservation, nor vice versa. You could perfectly
well have steady flow of a non-conserved quantity, with a steady rate of
production at one point and a steady rate of destruction at another point.

The programs used to create the video are given in reference 6.

6 Extensive Quantities

A quantity (such as charge) is considered to be extensive if you can divide the

system into a bunch of disjoint subsystems, and the charge on the overall
system is equal to the sum of the charges on the subsystems.

Extensive is not the same as conserved. There are four possibilities:

1. Some things are extensive but not conserved, as we can see from the
following example: Consider heat flow in a region where some heat is
being produced, perhaps by a tiny radioactive seed. The heat is still
extensive. It can be described by a continuity equation with a source
 term. (This stands in contrast to a conserved quantity, which is
described by a continuity equation without source/sink terms.)

2. Conversely, some things are conserved but not extensive, e.g. the energy
of a crystal. If you cleave it into tiny pieces you increase the surface
area. There is energy per unit area due to broken and/or reconstructed
bonds. For each crystallite, some of the energy scales like size cubed (i.e.
volume), but some of it scales like size squared (i.e. surface area).
See reference 7 for a discussion of what we mean by scaling.

3. Some things are neither extensive nor conserved. In figure 11, there is
one photon in box A, which has zero mass, plus one photon in box B,
which also has zero mass. However, the mass of the overall system AB is
nonzero. So mass is not extensive. Meanwhile, if the photon from
box B were to cross into box A, the mass within box A would change,
even though no mass crossed the boundary. So mass is not conserved.
For details of how to calculate the mass of such things, see reference 1.

Figure 11: Photon System and Subsystems

In high-school chemistry class it is often taken as an article of faith that mass
is an extensive quantity, sometimes even to the point where “extensive”
is defined as “proportional to mass”. That’s mostly OK in the context of
chemistry, because chemists don’t spend a lot of time weighing photons ... but
if you look closely enough, you find that it’s only an approximation. Mass is
not strictly conserved, and not strictly extensive.

4. Some things are both conserved exactly and extensive exactly (so far as
anybody knows), e.g. electric charge. Some things are conserved exactly
and extensive to a good approximation, e.g. the energy of a large system,
insofar as we can neglect boundary contributions.

Roughly speaking, the extensive property has to do with what happens when
you look at the system at a single moment in time and chop it up in spacelike
directions. In contrast, conservation has to do with how things flow and
change, which involves the time direction.

Conservation : The Basic Idea

When electric charge flows from place to place, it obeys a local conservation
law. By that we mean something very specific:

Any decrease in the amount of charge in a given region of space

must be exactly balanced by
a simultaneous increase in the amount of charge in an adjacent region of
space.

We can express the same idea in the form of an equation:

change in
= net flow of charge
charge
(1)
(inside (inward minus outward across
boundary) boundary)

The word “flow” in this expression has a very precise technical meaning,
closely corresponding to one of the meanings it has in everyday life.
Specifically: Charge cannot be created or destroyed, but it can flow one from
one region to another. Any decrease in one region is exactly balanced by a
simultaneous increase in an adjacent region.

(By way of contrast, if you want an example of something that flows but
is not conserved, consider “the flow of ideas”. You can give me an idea without
decreasing your own stock of ideas. As usual, flow refers to spreading from
one region to a neighboring region.)

At this point we have three ideas on the table: conservation, change, and flow.
In accordance with equation 1, we can understand conservation as a
combination of change plus just the right amount of flow.

To repeat: electric charge cannot change arbitrarily; only certain types of

changes are allowed. The relationship between the broad concept of “change”
and the more-precise concept of “conservation” can be summarized as follows:

Conservati chan equal-and-opposite change +

= +
on ge simultaneity + adjacency (
chan 2)
= +matching flow
ge

The use of the world “simultaneity” causes some people to wonder

whether equation 2 can be relativistically correct. Well, it turns out that this
equation is 100% kosher, relativistically and otherwise. We get into trouble
with relativity when we talk about simultaneity at a distance, but since we are
taking about adjacent changes – zero distance – there is no problem with
simultaneity. Indeed, our notion of conservation is beautifully matched to
modern notions of relativity. In spacetime, local conservation can be
expressed in elegant, informative, and easy-to-visualize form, in terms of
the continuity of world lines. This is discussed in section 2.

To be useful – or even correct – the conservation law must be local, as

discussed in section 2.3. It must apply to every region in spacetime, including
small open regions where the charge is not constant. Charge
is always conserved, even though it is not always constant, as discussed
in section 2.4.

As a corollary, as a special case of equation 2:

• If there is no flow across the boundary, conservation implies there

cannot be any change.

• If there is no flow across the boundary, conservation implies constancy.

This stands in contrast to the general case, where conservation does not imply
constancy, nor vice versa. This is discussed in section 2.4.

To summarize: The laws of physics do not permit charge to disappear now and
reappear later. Similarly the laws do not permit charge to disappear from here
and reappear at some distant place.

Charge is conserved right here,

right now.

2 Conservation in Spacetime

2.1 Conservation
An ordinary box in D=3 space has six faces. We can call them the ± x faces, the
± y faces, and the ± z faces. In D=4 spacetime, the corresponding object has
eight faces, namely two ± t faces in addition to the six spatial faces.

Now let’s draw some spacetime diagrams. For simplicity we will conceal
the y and z dimensions, and portray only the ± t and ± x faces of the box.

Figure 1 shows the world-line of a stationary parcel of electrical charge sitting

at the location of our box. The world-line enters on the t=0 face and exits on
the t=1 face. In other words, it was initially within the spatial extent of our
box, and it remained so.
Figure 1: World-line of Stationary Parcel

World-lines crossing the ± t faces contribute to the time derivative of the

amount of charge in the box. In figure 1, there are two equal and opposite
contributions, namely one flowing in at t=0, and one flowing out at t=1.

In 4-dimensional spacetime In 3-dimensional spatial terms:

terms: The charge is flowing in The charge is not flowing at all.
the time direction only. It just sits there. The amount of
charge within the spatial extent
of the box is the same at t=0
and t=1, so the amount of charge
in that spatial region is constant,
i.e. not changing with time.

We now consider what happens when the charge is non-stationary, as shown

in figure 2. The world-line enters via the t=0 face and exits via the x=1 face. In
other words, the parcel was initially within the spatial extent of our box, but
then it departed toward the right.
Figure 2: World-line of Moving Parcel

In this case, the amount of charge in the spatial box at t=1 is less than the
amount of charge at t=0. The charge is not locally constant, but it is locally
conserved. The charge wasn’t destroyed, it just flowed across the x=1
boundary into the neighboring region.
We will make this more quantitative in a moment. To make this useful, we will
need to quantify the net flow across the ± x boundaries. It helps to know
something about velocity and charge density.

You can see where this line of reasoning is leading. The following notions are
formally equivalent:

• Local conservation.

• Continuity world-lines.

2.2 Non-Conservation

We now consider a non-conserved object, such as a muon. Here is the world-

line of a muon that decays within the box; the world line enters the box but
doesn’t go out again:
Figure 3: World-line of Non-Conserved Object

Here are some trickier examples of non-conservation. In figure 4, the particle

disappears and then reappears at a later time. Electric charge does not behave
like this. The law of conservation of charge forbids it. In particular, if we
divide the region into an early half and a late half, we now have two regions,
both of which exhibit gross non-conservation. The lesson here is that the
conservation is not equivalent to saying that you end up with the same
amount you started with. The conservation law applies not just at the starting
time and ending time, but at every time along the way.

Figure 4: Non- Figure 5: Non-

Conservation: t Gap Conservation: x Gap

Similarly, in figure 5, the particle disappears at one location and reappears at

some other location. Electric charge does not behave like this. The law of
conservation of charge forbids it. In particular, if we divide the region into left
half and a right half, we now have two regions, both of which exhibit gross
non-conservation. The lesson here is that the conservation is not equivalent to
saying that the total amount is constant over time. The conservation law
applies not just to the universe as a whole, but also to every region and
subregion.
2.3 Locality

It must be emphasized that conservation of charge is a local conservation law.

The property of “locality” is crucial, for at least two reasons, as we shall see.

It is possible to have global conservation without local conservation, as shown

in figure 5. If we consider only the large region from x=0 to x=1, there is no
change in the charge inside, and no flow across the ±x boundaries, so we have
conservation in accordance with equation 1.

The first huge problem with global conservation is that it is not useful in
practice. If some charge disappears from your laboratory and reappears
somewhere far, far away, you have no way of knowing whether it is really
conserved. You have no way of checking. An unenforceable law is useless.

The second huge problem manifests itself when we consider a second

reference frame, moving relative to the first, in accordance with the laws of
special relativity. The situation is shown in figure 7. In this figure, the world
lines are exactly the same as in figure 6 (and also figure 5); only the reference
frame is changed. Whereas an ordinary rotation in the xy plane rotates
the x and y contour lines the same way, a spacetime rotation in the xt plane
tilts the x contours one way and the t contours the other way. For details,
see reference 1. In figure 7, the amount of tilt corresponds to a x^t^ frame that
is moving at half the speed of light (relative to the xt frame).

Figure 6: Non- Figure 7: Non-Conservation: x Gap,

Conservation: x Gap Boosted

The point of all this is that in the x^t^ frame, the particle disappears later than
time t^=0.5 and reappears earlier than that. Therefore even though the
amount of charge is globally constant relative to the xt frame, it is not
constant relative to the x^t^ frame. Since the laws of physics are required to
work in any reference frame, we see that any notion of non-local conservation
is dead on arrival. In the real world, i.e. in spacetime, non-local conservation is
almost no conservation at all.1

2.4 Conservation versus Constancy

It is important to distinguish the notion of constancy from the notion

of conservation:

• When we say some quantity q is constant, or equivalently unchanging, it

means simply that the amount of q in the region does not change. There
are various ways this could happen, either with or without
conservation, as spelled out in the table below.

• When we say some quantity q is conserved, it means that the amount

of q does not change except by flowing across the boundary of the
region. Conservation can happen with or without change, as spelled out
in the table below.

Imagine a fluid that has the same velocity and the same charge-density
everywhere. An example of such a situation is shown in figure 8.

Figure 8: Constancy via Replacement of Conserved Particles

In this case, the amount of charge in the box is constant, because whenever a
parcel of charge leaves via the x=0 face, an equivalent amount of charge is
entering via the x=1 face. Contrast this with figure 2, where the charge flowed
out and was not replaced. This non-replacement could occur because the
charge density was lower off to the left, and/or the velocity was lower off to the
left. So we see we need to be concerned with

• the charge density

• the velocity

• the left-to-right variation thereof.

A particularly interesting case of constancy with flow and replacement is

steady flow around a loop, as shown in figure 9. The diagram is subject to
periodic boundary conditions. That is, the two dashed worldlines are
the same worldline. Also the two dashed red boundaries are
the same boundary. Every time a particle flows out of one compartment it is
replaced by another particle flowing in.

Figure 9: Constancy and Conservation: Steady Flow Around a Loop

We have just seen constancy with conservation, via outflow compensated by

inflow. We now constrast that against constancy without conservation, via
annihilation and inflow. This shown in figure 10. You can see that the number
of objects in the region of interest is constant, because every time one of them
is destroyed it is replaced.

Figure 10: Constancy via Replacement of Non-Conserved Particles

A real-life example of this concerns the number of working light-bulbs in a

room. The number of working light bulbs is not a conserved quantity, but the
number is approximately constant, because every time a light bulb is
destroyed you replace it.
Another example along the same lines is a continuous-flow reactor in a
chemical plant, where chemicals of type A flow in and chemicals of type B flow
out.

To be explicit: Conservation does not imply constancy, nor vice versa. Change
does not imply non-conservation, nor vice versa. All four of the possibilities
have already been illustrated:

constant non-constant

conservation Figure 1 (static) Figure 2 (outflow)

Figure 8 (steady flow)

non-conservation Figure 10 (steady replacement) Figure 3 (annihilation)

As touched upon section 1, if you integrate the local conservation law over all
space, you can derive a global conservation law as a corollary. It is true that
global conservation is the same as global constancy ... but that does not mean
that conservation is the came as constancy in general.

Constancy can be used as a proxy for conservation for closed systems only. By
far the more powerful idea is conservation (and local conservation in
particular), because it can be applied to small, open sytems. In particular,
suppose we have a small region. Charge is flowing across the boundary.
Charge is conserved, even though the charge in our region is not constant. We
could try to salvage the idea of constancy by considering a larger region, but
then we lose the idea of locality ... and without locality we are doomed, for
multiple reasons, as discussed in section 2.3.
In summary: The only way to do the whole job properly is to define a notion of
conservation that works for small, open systems – that is, local conservation,
namely equation 1. The key ideas are:

• To be useful, the law must be local. Furthermore, special relativity says

that a non-local law cannot possibly be correct.

• Locality requires us to consider small systems.

• Small systems are generally not closed.

• Therefore a law that applies only to closed systems is nowhere near

strong enough to be a fundamental law of physics.

• The fact is, charge is always conserved ... even in small, open regions
where the charge is not constant. This is true and important.

3 Discussion

3.1 Many Are Conserved

We have been using charge as our primary example of a conserved quantity.

There is a long list of such things.
 • Charge is conserved.

• The [energy, momentum] 4-vector is conserved.

• If we decompose the [energy, momentum] 4-vector into components in
any chosen basis, the x, y, and z components of momentum are
separately conserved. This can be used to derive a useful equation of
motion for fluids, namely the Euler equation; see reference 2.

• Similarly, in any chosen basis, the energy is separately conserved. This

makes sense in relation to the previous item, given that the energy is
the fourth component of the [energy, momentum] 4-vector. Conservation
of energy is synonymous with the first law of thermodynamics, as
discussed in reference 3.

• Lepton number is conserved.

• Et cetera.

The picture of world-line continuity applies to them all.

In addition, there is an even longer list of things that are approximately

conserved.
• Entropy obeys an equation very similar to equation 1, but it is an
inequality rather than an equality.

• If we restrict attention to reversible processes, entropy obeys a

conservation law. (This is not an exact law of nature, insofar as nothing
is exactly reversible.

• The mass of a particle is not conserved.

Mass is a relativistic invariant, i.e. a Lorentz scalar, related to the invariant

norm of the [energy, momentum] 4-vector. For a particle with nonzero mass,
this is proportional to the rest energy in accordance with the celebrated
formula Erest = m c2. For a slow-moving massive particle, the energy is very
nearly equal to the rest energy, which implies that the mass is conserved to a
good approximation (but not exactly).

• In a chemical reaction, the number of atoms of a given type is

conserved. (This is what distinguishes chemical reactions from other
things such as nuclear reactions.) If there are 100 different kinds of
atoms involved, this gives us 100 separate conservation laws, one for
each atom ... in addition to all the non-chemical conservation laws
(energy, charge, et cetera).

3.2 Warning: Conflicting Terminology

Beware: The term “conservation” has two wildly different meanings.

The physics notion of The homespun notion of

conservation is the topic of this conservation refers to protecting
document. It is defined by
 expressions of the something against destruction,
form equation 1. or especially against waste.

Example: Charge is always Example: Conservation of

conserved – automatically, endangered wildlife. People get
strictly, and locally – in to decide whether they want to
accordance with equation 1. conserve this-or-that. Sometimes
they decide yes, sometimes no.

While we’re on the subject: Beware: There are two different notions of energy.
There exists a homespun notion of «useful energy» that is important and
widely used, yet is very different from the physics energy.

The physics energy includes all The homespun energy refers to

forms of energy, whether useful some ill-defined notion of “useful
or not. energy” or “available energy”.

Example: The rest energy is Examples: The energy industry.

given by E=mc2. The Department of Energy.

Suppose you see a statement of the form “Please conserve energy by turning
out the lights”. Be careful not to intepret that in physics terms, because both
of the key words – “conserve” and “energy” – are being given meanings wildly
different from their physics meanings.

3.3 The Current is Not Conserved

Just because electrical charge is conserved does not mean that electrical
current is conserved. In fact, the current is definitely not conserved. Imaging a
loop of wire with some current flowing in it initially. If you just leave it alone,
the current will die out. The current will not move into some neighboring
region; it will just remain in place and get weaker. This is an absolutely clear
example of non-conservation.
This is mostly just a terminology issue, because physically the current is
intimately associated with the thing that is conserved. Indeed, you cannot
even write down the equation for conservation of charge without mentioning
the current also:

∂
ρ + ∇·
0 = (3)
j
∂t

We say that the current obeys an approximate continuity equation (not a

conservation equation, approximate or otherwise). More specifically, in the
low-frequency limit – for so-called “DC” circuits – we can ignore the first term
on the RHS of equation 3, and what’s left is a continuity equation for the
current, namely equation 4:

∇· (3D continuity of DC
0≈ (4)
j current)
 (implies no
accumulation)

One of Kirchhoff’s «laws» says that current obeys a continuity equation.

People very often design circuits so that they uphold Kirchoff’s laws, to a good
approximation ... but this is a choice, not a law of nature, as discussed
in reference 4.

While we’re in the neighborhood, if we instead ignore the other term on the
RHS of equation 3, we get:

∂
ρ (constancy of
0 ≈ (5)
charge)
∂t

Zero accumulation does not imply zero flow. It just means no net flow across
the boundaries. You can have lots of flow-with-replacement. You can also have
lots of flow going round and round in the interior, not crossing the
boundaries.

In fluid dynamics, vortex lines obey a continuity equation. In electrodynamics,

magnetic field lines obey an exact continuity equation in three dimensions. In
any region where there is no electrical charge, electric field lines obey a
continuity equation.

The important idea is that the full conservation equation – equation 3 – can be
understood as a spacetime continuity equation, as discussed in section 2.

In some weird sense, the idea of constancy (equation 5) is an amputated part

of the idea of conservation, and also the idea of three-dimensional continuity
(equation 4) is an amputated part of conservation. These two terms working
together give the proper idea of conservation, i.e. four-dimensional continuity
(equation 3).

Bottom line: You shouldn’t talk about “conservation” of the current vector;
that gives entirely the wrong idea. If you want to say something about the
current, talk about the spacetime continuity of the [charge, current] 4-vector.

4 Quantifying Conservation

Now that we have a firm qualitative idea of conservation, let us now make the
idea completely quantitative.

4.1 Differential Version

For an electrically-charged fluid, the product of charge density times velocity

is called the current. A spatially-uniform current isn’t interesting from a
conservation point of view, so we shall be particular interested in the spatial
variation in the current. If there is more than one spatial direction, the
current is a vector (just like the velocity) with x, y, and z components. What
matters is the x-derivative of the x-component of flux, the y-derivative of the y-
component, and the z-derivative of the z-component — in other words, the
divergence.

So we can write the local conservation law as

∂
(charge density) = − div(current density) (6)
∂
t

or equivalently

∂
ρ + ∇·j = 0 (7)
∂
t

This is a 100% formal accurate statement of the local conservation law. It

expresses the continuity of the world-line of each little bit of charge.

Homework: Check equation 7 using dimensional analysis. It shouldn’t take

very long!

The structure of equation 7 is very elegant. It has the form

∂ ∂ ∂ ∂
(...) (...) (...) (... ..
= (8)
∂ + ∂ + ∂ + ∂ ) .
t x y z

which is most elegantly understood as a four-dimensional divergence. If you

leave off the ∂/∂t term you’ve got the plain old three-dimensional divergence.
The key concept is that the world lines of a conserved quantity (such as
charge) are divergence-free in D=4. They have no beginning or end. By way of
contrast, if you projected the world lines onto three dimensions, they would
not be divergence-free; there would be a nonzero divergence at any location
where charge is accumulating. The timelike term in equation 8 is needed to
make the equation correct. On the third hand, we can contrast this with
magnetic field lines, which are divergence-free – i.e. endless – even in three
dimensions.

In more detail: The world line of an element of a charge has a tangent vector
at every point. This defines a vector field. At each point along the world line,
there is a four-vector consisting of the three components of current plus a
fourth component, representing the “flux” of charge flowing in the “time
direction” – which is just the density of charge itself. Overall, this vector field
is divergence-free in D=4.
Equation 7 is frame-independent. Each of the terms on the LHS is frame-
dependent, but the LHS as a whole is the same in all frames, the same for all
observers. The RHS is zero, which is manifestly invariant.
To repeat: In D=3 you can (temporarily) have a whole bunch of current
diverging from some region of space. [In contrast, you cannot have magnetic
flux lines diverging like this, not even temporarily.] The divergent current will
deplete the charge density in that region. Meanwhile, if you take a D=4 view of
the same situation, you will find that the [charge, current] four-vector has zero
divergence. The world-lines of any conserved quantity are endless.

In the preceding paragraph, all the D=4 statements are frame-independent, i.e.
Lorentz invariant. (As always, any D=3 statements are necessarily not Lorentz
invariant.)

The formulation in this section applies at a single point, or perhaps in the

infinitesimal neighborhood of a point. (A neighborhood is necessary to make
derivatives meaningful.) It involved density and flux, in particular the four-
dimensional divergence of the [charge-density, current-density] 4-vector.

Similar laws apply to the [energy, momentum] 4-vector, but the details are
messier. Note the contrast:

• For charge, the conserved quantity is charge itself. It is a scalar. The

current is a vector that is “associated” with the charge, but the current
is not per se the conserved quantity.

• For energy, things are quite a bit more involved.

Let’s choose a particular basis, and project out the components of the [energy,
momentum] 4-vector. We start by focusing attention on the energy component.
This is a conserved quantity, in much the same way that charge is a conserved
quantity. There is a slight difference, because charge is a Lorentz invariant
whereas energy is not, but let’s pretend we didn’t notice that. In any particular
chosen basis, that doesn’t matter very much.
Following the pattern of equation 8, we need four terms to express
conservation of energy: One for the time-dependence of the energy, and three
more for the divergence of the flow of energy.

Similarly, we need four terms to express conservation of the x-component of

momentum, and four more for the y-component of momentum, and four more
for the z-component of momentum.
Altogether, that means we need sixteen terms to express conservation of the
[energy, momentum] 4-vector. These can be arranged in a 4×4 array, i.e. a
second-rank tensor, namely the stress-energy tensor.

It takes a bit of work to keep track of how momentum is conserved, but it is

doable. The stress includes pressure, tension, and shear.

o Pressure involves things like x-momentum flowing in the

+x direction.

o Tension involves things like x-momentum flowing in the

−x direction.
 o Shear involves things like the x-momentum flowing in the y-
direction, and z-momentum flowing in the x-direction, et cetera.

For more information about this, including suggestions on how to diagram the
flow of momentum, see reference 5.

4.2 Integral Version

Rather than considering a single point, we now consider a region.

We take the differential version (as discussed in the previous section) and
integrate it over a region. Using Stokes’s law, we find that the time-rate-of-
change of current inside the boundary is equal to the amount of current
flowing across the boundary.

The integral formulation implies the differential formulation and vice versa,
subject to mild restrictions. Obviously things must be sufficiently
differentiable, and the region of interest must be simply connected (no
wormholes).

5 Conservation and Continuity Without Constant Density

Consider the flow pattern shown in the following video. Charged particles
cross the boundary into our field of view at the left, flow from left to right, and
then exit our field of view. At no point are any particles created or destroyed.
There is strict conservation of charge and continuity of current.
Furthermore, this is a steady flow – not just conserved, but also steady. Steady
flow means the amount of charge crossing any given contour of constant x is
the same, independent of time, for all times between the beginning and end of
the movie. (If you imagine that the end of the movie portrays the charged
particles coming to a stop, that is an example of unsteady flow. Charge is still
conserved, but the flow is unsteady.)

Thirdly, you can see that the current is independent of x. This can be
understood as a corollary of the fact that the flow is both conservative and
steady.

Time 0

However, this does not mean that charge behaves like an incompressible fluid.
We have conservation of charge, continuity of current, and steady flow, but all
of that together is not sufficient to guarantee that the density of charge
remains constant.

Let’s be clear: constant density would be one way of guaranteeing that what
goes in must come out ... but it is by no means the only way, as you can see in
the video. Conservation of charge (and continuity of current) just means that
if the density decreases, the speed must increase, so that the current remains
the same.
Situations like what you see in the video happen all the time in the real world.
For example,
 • This happens whenever you make a junction between two different
types of metal; the carrier density is likely to be substantially different
in the two metals.

• The same is true to an even more extreme degree at the junction

between a metal and a semiconductor.

• You see the same sort of thing on the highway. Cars are conserved, in
the short term, to a decent approximation. When you come to a zone
with a high speed limit and good driving conditions, the cars spread out
and speed up. When they get to a zone with a lower speed limit and/or
worse conditions, they slow down and get closer together.

Again: Please do not imagine that charge behaves like an incompressible fluid.
A lot of people who ought to know better get this wrong. The fact is, in the
video, the density in the middle is a factor of two less than the density near the
ends.

Here’s another important distinction: If we watch at any fixed location along

the x axis, the density at that location is independent of time. In contrast, if we
follow a particular parcel of fluid as it flows along, the density of that parcel
particle changes dramatically. So if you ask whether this is a “constant
density” situation, the answer very much depends on which viewpoint you
take.

In particular, if you want Kirchhoff’s laws to be reliable, you need to restrict

attention to the DC limit, i.e. to steady situations. (In other situations, you can
engineer things so that the violations of Kirchhoff’s laws are small, but this is
an engineering choice, not a law of physics.) For details on this,
see reference 4. In such a situation, we know the charge density at any
particular point is constant because the flow is steady ... certainly not because
it is allegedly incompressible. It’s actually quite compressible.

When you see look at some chosen window of observation, constant density
does not tell you much about conservation, nor vice versa. You could perfectly
well have steady flow of a non-conserved quantity, with a steady rate of
production at one point and a steady rate of destruction at another point.

The programs used to create the video are given in reference 6.

6 Extensive Quantities

A quantity (such as charge) is considered to be extensive if you can divide the

system into a bunch of disjoint subsystems, and the charge on the overall
system is equal to the sum of the charges on the subsystems.

Extensive is not the same as conserved. There are four possibilities:

1. Some things are extensive but not conserved, as we can see from the
following example: Consider heat flow in a region where some heat is
being produced, perhaps by a tiny radioactive seed. The heat is still
extensive. It can be described by a continuity equation with a source
 term. (This stands in contrast to a conserved quantity, which is
described by a continuity equation without source/sink terms.)

2. Conversely, some things are conserved but not extensive, e.g. the energy
of a crystal. If you cleave it into tiny pieces you increase the surface
area. There is energy per unit area due to broken and/or reconstructed
bonds. For each crystallite, some of the energy scales like size cubed (i.e.
volume), but some of it scales like size squared (i.e. surface area).
See reference 7 for a discussion of what we mean by scaling.

3. Some things are neither extensive nor conserved. In figure 11, there is
one photon in box A, which has zero mass, plus one photon in box B,
which also has zero mass. However, the mass of the overall system AB is
nonzero. So mass is not extensive. Meanwhile, if the photon from
box B were to cross into box A, the mass within box A would change,
even though no mass crossed the boundary. So mass is not conserved.
For details of how to calculate the mass of such things, see reference 1.

Figure 11: Photon System and Subsystems

In high-school chemistry class it is often taken as an article of faith that mass
is an extensive quantity, sometimes even to the point where “extensive”
is defined as “proportional to mass”. That’s mostly OK in the context of
chemistry, because chemists don’t spend a lot of time weighing photons ... but
if you look closely enough, you find that it’s only an approximation. Mass is
not strictly conserved, and not strictly extensive.

4. Some things are both conserved exactly and extensive exactly (so far as
anybody knows), e.g. electric charge. Some things are conserved exactly
and extensive to a good approximation, e.g. the energy of a large system,
insofar as we can neglect boundary contributions.

Roughly speaking, the extensive property has to do with what happens when
you look at the system at a single moment in time and chop it up in spacelike
directions. In contrast, conservation has to do with how things flow and
change, which involves the time direction.