BBC has a report of a region that seems to be a galaxy without stars. Says here in the dictionary that the word galaxy comes from a Greek word for milk, so the phrasing might seem a bit odd. What they actually found is a huge amount of hydrogen atoms, which appears to be rotating; and rotating at a speed greater than could be accounted for by merely the mass of the hydrogen. Normal galaxies seem to have the same problem, hence one of the arguments for dark matter..
When excited by other radiation, hydrogen atoms emit photons with distinctive energies, and from the spread of relative red shifts of a cloud you could tell if a cloud of gas was rotating or not. The abstract says that they think it might be "part of a currently infalling population," but that "This observed column density is above the normally expected level for star formation to occur. The two detections with no optical counterparts have very much lower column densities than that of the rest of the sample, below the star formation threshold."
Or to translate that, they have two candidates for "galaxies that haven't quite formed yet." Or for galaxies that are formed mainly of "dark matter."
Given the fractions being bandied about for how much of the universe is supposed to be dark/non-baryonic matter, it wouldn't be crazy to expect to find some galaxies with less-than normal amounts of matter. To actually predict how low you can go in the amount of hydrogen (and thus of stars) in a galaxy I'd have to model this and run some fairly complex fluid-dynamic code; and I don't have the time for it. I'd be interested in seeing how this pans out.