Is speed an intensive property? I remember being taught in elementary physics that while it makes
sense to add volumes, masses, or heat, it makes no sense to add
temperatures.
As I wanted to use that to illustate some other issue, I checked it on
wikipedia, and discovered the concept of intensive and extensive
properties of materials and systems, which is apparently a century
old, though I do not remember it being taught to me (it was not as old
then :-).
The concept seems to be particularly useful in material and systems
thermodynamics. But I was wondering how far it extends, and I was
keeping in mind my original problem of determining when adding values in
a given unit (not necessarily a physical one) made sense or did not.
I noticed the fact that the ratio of two extensive variables is
intensive. So I started looking, a bit ramdomly I confess, at ratios,
and the first that came to mind was speed. And I wondered whether
adding speeds made sense.
We do that all the time, so it should. But then, my initial problem
was not about adding two quantities, but a long list of them (it was a
database question). While adding a few speeds (or velocities) makes sense
when I move in the bus, or analyse the motion of the Moon in the solar
system, I cannot imagine it would ever make sense to add a list of a
hundred or a thousand speeds (though computing the average would make
sense, as it would for temperature). But I makes perfect sense to sum
the masses of a thousand objects.
So there is apparently something special, not quite right, about
adding speeds.  Of course, I know that slightly older results state
that adding speeds is not done with simple addition, but I feel that my
problem is elsewhere  (though there may possibly be a connection).
I am aware that temperature in some materials is related to speed of
motions inside it, so I am not too surprised. But speed in general
seems to go beyond that (sorry for the vagueness). Also speed is not
listed by Wikipedia as an intensive property.
So, my question is : why does it seem improper to add many speeds (or velocities)?
I guess this must also be true of other physical quantities, and I am
wondering what is the right way to look at this, and understand it. Is
there a more general notion than intensive and extensive?
 A: 
why does it seem improper to add many speeds (or velocities)?

Adding speeds is ofttimes inappropriate even in Newtonian mechanics. Suppose Mark is moving 3 m/s eastward with respect to Bob, and John is moving 3 m/s westward with respect to Mark. The relative velocity between Bob and John is zero rather than the 6 m/s suggested by adding speeds. You can add velocities however in this case.
Velocity however is neither an intensive nor extensive property. Consider a big brick moving at $\vec v_b$ and a grain of sand moving at $\vec v_s$. What is the velocity of the brick+grain system? Typically one would use the velocity of the center of mass, and that's $\frac {m_b \vec v_b + m_g \vec v_g}{m_b + m_g}$. That expression doesn't correspond to either the concept of an intensive property or an extensive property.
Momentum rather than velocity is additive, at least in Newtonian mechanics. The momentum of the brick+grain system is the vector sum of the momentum of the brick and the momentum of the grain of sand. Thus momentum (but not velocity or speed) is an extensive property in the context of Newtonian mechanics.
A qualification is needed here: Momentum is additive in Newtonian mechanics if those momenta are all referenced with respect to a common frame of reference. The momentum of the brick with respect to some observer and the momentum of the grain of sand with respect to the brick don't add up.
I've used "Newtonian mechanics" a number of times. The Newtonian concept of momentum as the product of mass and velocity fails when things are moving very fast. In the context of special relativity, it's relativistic momentum rather than Newtonian momentum that is additive (and once again, the concept of a common frame of reference very much applies).
In the context of general relativity, even adding relativistic momenta doesn't quite make sense. Asking about the total momentum of the Andromeda galaxy and an extremely remote galaxy is a bit nonsensical in the context of general relativity.
A: Your question, as of right now, seems confused to me. An extensive property of a system is one that scales with the system size. An intensive property is independent of the system size. For example, consider a system $A_1$ with $N$ particles in a volume $V$, with density $\rho=\frac{N}{V}$. Now, we consider two of these systems separately, $A_1$ and $A_2$, and call the two of them  together $B$. Then, $B$ has a total number of $2N$ particles, within a volume $2V$, with density $\rho=\frac{2N}{2V}=\frac{N}{V}$. 
As we see, 'doubling the system' gives us twice the number of particles and twice the volume, but the same density. This shows that particle number & volume are extensive quantities, and density is an intensive quantity. Now, think of a particle in a box bouncing around at velocity $v$. Consider two such systems, and you will have two boxes with a particle, each with velocity $v$. Clearly, the velocity of each particle is unaffected by our silly thought experiment, so it is clearly not an extensive property.
Can we now say that the particle velocity is an intensive property? Not really. The notions of intensive- and extensiveness arise in thermodynamics, where they are useful in grouping certain properties of systems together based on whether or not they scale with system size. In thermodynamics, one usually considers systems which are a (large) collection of individual particles, so that one can discuss things like temperature and other notions that don't really make sense on a microscopic level. 
If one considers a large collection of particles, then the velocity of a single particle really isn't a property of the system. The whole point of thermodynamics is to understand the behavior of large systems without discussing the details of what's going on with each single constituent. Therefore, the velocity of a single particle is not usually considered a property of the system to begin with, so it cannot be either intensive or extensive. I hope this clears things up a little.
However, in other contexts it of course makes perfect sense to add velocities, e.g. in Newtonian mechanics or special relativity theory. 
A: Taking answers and comments into account, my own current conclusion is that velocity is an intensive property, provided the system considered is homogeneous, at least with respect to speed. Like other intensive properties, this may depend on scale, and cease to have meaning at molecular level.
I did not intend to write an answer to my own question but ... writing
the question is usually a good way to better understand the issue and
to find the answer. I could have added this to the question, as there
are remaining issues at the end. But the question is already long, and this is long too. Hence it seemed better to do it this way.
I initially looked at speed as a ratio because I know it as the ratio of a
distance to a time.  But that was much without thinking, and is the wrong way to look at it,
especially since I do not know too much what distance and time may be. I actually stayed with speed, or rather velocity, when I realized I would not see the meaning of adding many speeds, which was  a variant of my original database problem.
But then, I realized that velocity is a ratio in a
different way: $\overrightarrow{V}=\overrightarrow{P}/M$, i.e.,
velocity is the ratio of the momentum (vector) by the mass.
And then, David Hammen's answer timely reminded me that momentum is
additive in Newtonian mechanics, when considering the motion of
several bodies in the same frame, and also that velocity is not ... which
confirms that, if anything, it is not extensive.
Indeed, if I take the answer of Nathaniel to another question, as well
as the definition of the IUPAC Green Book, recalled in the wikipedia
page, additivity is what characterizes an extensive property. I
must then conclude that (linear) momentum is an extensive property,
and velocity is not.
The same reasonning and conclusion applies to mass (this seems
generally agreed upon :-).
Since velocity is the ratio of the momentum vector by the mass, it
ensues that velocity must be an intensive property of a system.
This justifies that velocity (or speed) is not additive, in the sense
that you cannot make a large system from smaller parts, and add the
velocities of the parts to get the velocity of the whole.
But this conclusion is in contradiction with David Hammen's view that
velocity is not intensive, though he does not really say why, and I do
not see why his formula for the velocity of the center of mass should
be a problem.
However, I think that a requirement (stated only in examples, but not
in definitions in the documents I have) is that intensive properties
are meaningful only in homogeneous systems in equilibrium (which was
just confirmed by the comment of CuriousOne. With regard to velocity,
that would mean that it can apply only when the system considered is velocity-homogeneous. or
somehow merge into a single homogenous system. Maybe it could apply to
the collision of gas clouds in space, or of water blobs in zero
gravity (in a space ship). There are probably other examples of fluids
that can merge and balance momentum to reach equilibrium, without
breaking apart.
That should make velocity an intensive property.
Velocities may be added when considering relative velocities of different
systems with respect to each other. But that is a completely different
situation, and it has nothing to do with merging
several systems into a single system, and properly "combining" their velocities. It is unlikely to lead anyone to add together a large
number of velocities, as one might add many momenta when having a
large number of bodies merged into a single system.
In the above discussion I have been considering blobs or clouds of
fluid as exemple, as that seem necessary to be able to talk of merging
two systems. However, there is no need for such restriction when just
considering a single system, say a stone or a cube of iron. If
velocity as an intensive property makes sense at all, it does make
sense for any velocity-homogeneous, and that certainly
applies to a solid.
When considering relativistic velocities, there is no reason the question should
lose its meaning.
A supernova can expel clouds of material at 10% of the speed of light,
which may be considered a relativistic speed as it corresponds to a 1%
discrepency between Newtonian and relativistic mechanics. I guess that
such a cloud, or part of it, could hit and other cloud and merge with
it into an ultimately homogeneous cloud (but I have zero expertise on
such phenomena). The penultimate paragraph of David Hammen's answer,
as well as the comment by Void, seem to indicate that things still
work at relativistic speed, provided the right properties are used
(such as relativistic momentum according to David Hammen, or four-momentum
and four-velocity according to Void).).
Finally, a supported comment by leftaroundabout suggests that "intensive vs. extensive only
makes any sense for true scalar quantities", but without further
justification. I am wondering why that should be. Besides, this seems
in disagreement with the Wikipedia page that lists magnetization as an
intensive property. But magnetization is a vector field.
Similarly, I do understand Danu's answer that the concepts of
intensive and extensive properties were developed as useful tools for
studying thermodynamics (I did read a bit before asking my
question). But according to the wikipedia page, they are already used
in other contexts, and I do not see why that should exclude
considering velocity, when statistically as homogeneous, as I explained above.
A: Here's a point of view from thermodynamics that might be useful.
Typically, the intensive quantities (in the form they're usually defined) arise as derivatives of the total (internal) energy $U$ by some particular extensive quantity. Thus:


*

*Temperature $T=\frac{\partial U}{\partial S}$, the derivative with respect to the entropy

*Pressure $P=-\frac{\partial U}{\partial V}$, the derivative with respect to the volume (with the opposite sign convention to the other quantities, but this isn't important)

*Chemical potential $\mu_i=\frac{\partial U}{\partial N_i}$, the derivative with respect to particle number of a given species

*Electrical potential $\phi=\frac{\partial U}{\partial Q}$, the derivative with respect to charge
and so on. These can be summarised as the "fundamental equation of thermodynamics:
$$
 dU = TdS - PdV + \sum_i \mu_i dN_i + \dots
$$
Now, the question is, can we add a term along these lines that involves speed, or rather velocity? The answer is yes, because
$$
v_x = \frac{\partial U}{\partial p_x},
$$
where $v_x$ is component of a particle's velocity in the $x$ direction, and $p_x$ the $x$ component of its momentum. Similar equations apply for the $y$ and $z$ directions of course.
The above equation is quite general, and applies in relativistic situations. However, if we restrict ourselves to a simple Newtonian particle in one dimension, for which $U = V(x) + \frac{1}{2}mv^2$ (where $V(x)$ is the potential energy) then it's easy to see that it's true in this special case, since we may rewrite the energy as $V(x) + \frac{1}{2m}p^2$.
Therefore, if the situation warrants it, we can add a $vdp$ term to the fundamental equation of thermodynamics, and thus velocity (in some direction) may be seen as the intensive quantity corresponding to the momentum in the same direction, which is extensive. (Note that it does make sense to add, rather than average, the momenta of thousands of particles.)
In fact this is often done when applying non-equilibrium thermodynamics to fluid mechanics, and to atmospheric dynamics in particular. The difference in horizontal velocities between the prevailing wind and the stationary Earth surface may be seen as the driving force that causes a flux of (horizontal) momentum downwards from the atmosphere into the Earth, leading to a tiny amount of frictional heating. This is exactly analogous to the way in which a difference in temperatures drives a flow of heat, or a difference in electric potentials drives a flow of electrons.
A: This is a very interesting question.  Velocity can be considered as either an intensive or an extensive property, depending on whether we are inquiring about the parts of a single system, or considering relations among separate systems. 
Velocity must be an intensive property, for consider: If I and my passenger and my books are traveling in my car, and if my car has the property of moving at 60 mph relative to an observer resting at the side of the road, then not only do I share in that property, but so does my passenger, every single one of my books, and every elementary particle in that car.  Hence, the velocity of the car is an intensive property with respect to all of the parts of my car.
The velocity of the car will not change if I throw one of my books out the window, although its momentum will change- especially if it happens to be a hefty book, such as J. A. Wheeler's Gravitation.  Whereas velocity is intensive, momentum is extensive.
The initial question seems to address the problem of adding the velocities of separate systems (of passengers, books, particles, etc.)  Suppose that I toss my book out the window of my car (doing 60 mph), and it happens to land in the bed of a truck speeding past me in the same direction at 40 mph.  Now my book is going 100 mph with respect to the rest observer.  If my hapless book then bounces out of the bed of the truck, and lands on a train going 180 mph (with respect to the rest observer) in the opposite direction, then my book is traveling -240 mph relative to me and my passenger.
We can watch my book bounce among distinct moving systems moving at different velocities (along the same dimension with respect to the rest observer) all day long.  So velocity is an extensive property with respect to separate systems.
Now here's another interesting question: how does the velocity of light serve to limit the addition of speeds of separate systems?
