thermodynamics - What is the speed of sound in space?

Given that space is not a perfect vacuum, what is the speed of sound therein? Google was not very helpful in this regard, as the only answer I found was $300\,{\rm km}\,{\rm s}^{-1}$, from Astronomy Cafe, which is not a source I'd be willing to cite.

Answer

By popular demand (considering two to be popular — thanks @Rod Vance and @Love Learning), I'll expand a bit on my comment to @Kieran Hunt's answer:

Thermal equilibrium

As I said in the comment, the notion of sound in space plays a very significant role in cosmology: When the Universe was very young, dark matter, normal ("baryonic") matter, and light (photons) was in thermal equilibrium, i.e. they shared the same (average) energy per particle, or temperature. This temperature was so high, that neutral atoms couldn't form; any electron caught by a proton would soon be knocked off by a photon (or another particle). The photons themselves couldn't travel very far, before hitting a free electron.

Speed of sound in the primordial soup

Everything was very smooth, no galaxies or anything like that had formed. Stuff was still slightly clumpy, though, and the clumps grew in size due to gravity. But as a clump grows, pressure from baryons and photons increase, counteracting the collapse, and pushing baryons and photons outwards, while the dark matter tends to stay at the center of the overdensity, since it doesn't care about pressure. This creates oscillations, or sound waves with tremendously long wavelengths.

For a photon gas, the speed of sound is $$ \begin{array}{rcl} c_\mathrm{s} & = & \sqrt{p/\rho} \\ & = & \sqrt{c^2/3} \\ & \simeq & 0.58c, \end{array} $$ where $c$ is the speed of light, and $p$ and $\rho$ are the pressure and density of the gas. In other words, the speed of sound at that time was more than half the speed of light (for high temperatures there is a small correction to this of order $10^{-5}$; Partovi 1994).

In a non-relativistic medium, the speed of sound is $c_\mathrm{s} = \sqrt{\partial p / \partial \rho}$, which for an ideal gas reduces to the formula given by @Kieran Hunt. Although in outer space both $p$ and $\rho$ are extremely small, there $are$ particles and hence it odes make sense to talk about speed of sound in space. Depending on the environment, it typically evaluates to many kilometers per second (i.e. much higher than on Earth, but much, much smaller than in the early Universe).

Recombination and decoupling

As the Universe expanded, it gradually cooled down. At an age of roughly 200,000 years it had reached a temperature of ~4000 K, and protons and electrons started being able to combine to form neutral atoms without immediately being ionized again. This is called the "Epoch of Recombination", though they hadn't previously been combined.

At ~380,000 years, when the temperature was ~3000 K, most of the Universe was neutral. With the free electrons gone, photons could now stream freely, diffusing away and relieving the overdensity of its pressure. The photons are said to decouple from the baryons.

Cosmic microwave background

The radiation that decoupled has ever since redshifted due to the expansion of the Universe, and since the Universe has now expanded ~1100 times, we see the light (called the cosmic microwave background, or CMB) not with a temperature of 3000 K (which was the temperature of the Universe at the time of decoupling), but a temperature of (3000 K)/1100 = 2.73 K, which is the temperature that @Kieran Hunt refers to in his answer.

Baryon acoustic oscillations

These overdensities, or baryon acoustic oscillations (BAOs), exist on much larger scales than galaxies, but galaxies tend to clump on these scales, which has ever since expanded and now has a characteristic scale of ~100 $h^{-1}$Mpc, or 465 million lightyears. Measuring how the inter-clump distance change with time provides a way of understanding the expansion history, and acceleration, of the Universe, independent of other methods such as supernovae and the CMB. And beautifully, the methods all agree.