A Dark Galaxy in the Virgo Cluster?

While updating my work website recently, I came across an old page on a dark galaxy candidate, VIRGOHI 21, that I had worked on back in 2004 - 2007. The page had not been updated since then, despite things having changed in the meantime. While updating it, I decided to write this blog post, giving some personal recollections and an explanation of my current thinking on the subject.

In 2005, I was first author on a paper about VIRGOHI 21 called “A Dark Hydrogen Cloud in the Virgo Cluster”. We had found the 'dark cloud' in a neutral hydrogen survey of the Virgo Cluster carried out a few years earlier with the Lovell Telescope at Jodrell Bank, and had already published initial results in an paper by Jon Davies the year before, “A multibeam HI survey of the Virgo cluster - two isolated HI clouds?”. The second cloud, VIRGOHI 27, was observed with the Giant Metrewave Radio Telescope in India and found to be a very faint galaxy, but deep optical images of the area of VIRGOHI 21 from the Isaac Newton Telescope in the Canary Islands did not show anything. Another oddity was that VIRGOHI 21 looked As if it was, like most galaxies, rotating - in observations with Arecibo we could see that the velocity changed from north to south. This made it look like a galaxy without any stars - a dark galaxy!

Isaac Newton Telescope image from our press release, the ellipse shows the extent of VIRGOHI 21 based on observations with Arecibo

We considered whether VIRGOHI 21 could be tidal debris. At that time, it was generally thought that the only way to form long streams was through slow, tidal interactions. We were able to rule this out as there was no large galaxy in the right position to have pulled VIRGOHI 21 into the shape we saw.

Sloan Digital Sky Survey image of VIRGOHI 21 from their Image of the Week gallery. The orginal caption reads: Radio telescopes at Arecibo and Jodrell Bank Observatory detect a large cloud of hydrogen gas at the center of the region of the sky covered by this image, but no corresponding objects can be seen in it. The rotation of this cloud indicates the presence of a significant mass of dark matter (matter that we cannot currently detect directly) as well.

The discovery led to headlines such as "Astronomers claim first 'dark galaxy' find" (New Scientist), "Astronomers find star-less galaxy" (BBC) and "Not even a twinkle out of galaxy with no stars" (The Times). VIRGOHI 21 even got its own entry in Wikipedia. But the story was far from over...

At the time the first paper was published, we were already planning high-resolution observations with the Westerbork Synthesis Radio Telescope in the Netherlands. We hadn't been able to detect VIRGOHI 21 with the Indian telescope earlier, but this time we saw it. These observations showed that VIRGOHI 21 was linked to a nearby galaxy, NGC 4254 by a bridge of hydrogen. This galaxy has an unusual, lopsided structure, with one very large spiral arm, and we had already discussed internally whether this could be linked to VIRGOHI 21 and dismissed the idea as something we did not have enough evidence to speculate on. Now we had the evidence.

We also had even deeper optical imaging from the Hubble Space Telescope that still failed to reveal any visible galaxy. The Hubble observations also answered an alternative scenario that some simulations suggested - a fast encounter that would rip gas out of NGC 4254 but would also, necessarily, pluck stars out of that galaxy and leave them floating freely in space in the same area as the gas. Had they been there, these stars would have been visible to Hubble - but we saw no evidence of them.

A three-colour (i, r, and B bands) image of VIRGOHI 21 from the Isaac Newton Telescope overlaid with contours of neutral hydrogen density (green) from observations with the Westerbork Synthesis Radio Telescope. The extent of the galaxy is remarkably similar to the ellipse from the Arecibo data (above).

We published our Westerbork and Hubble results in a paper called “21-cm synthesis observations of VIRGOHI 21 - a possible dark galaxy in the Virgo Cluster”. At about the same time, the ALFALFA team published their map of VIRGOHI 21, which showed that the neutral hydrogen stream extended further to the north. This wasn't particularly shocking - seeing a 'leading arm' in front of an interacting galaxy is not uncommon.

Animation of the neutral hydrogen data cube from Westerbork. Up indicates increasing recessional velocity. VIRGOHI 21 is the structure in the centre, connected by a bridge to NGC 4254 (near the top of the image). The ring-like structure near the bottom of the image is NGC 4262.

Both the ALFALFA data and our Westerbork data were then used by a team in France who were modelling the system as a 'hyperbolic' interaction, a kind of cosmic 'hit and run' where another galaxy shot past NGC 4254 very quickly and then left the area before it could be identified. It may seem surprising that we gave our data freely to someone who were trying to prove us wrong - but that's the way science works!

The French team found that: “High-speed collisions, although current in clusters of galaxies, have long been neglected, as they are believed to cause little damages to galaxies except when they are repeated, a process called ‘harassment.’ In fact, they are able to produce faint but extended gaseous tails.” In other words, it was possible to explain VIRGOHI 21 as part of a tidal tail, formed from a high-speed galaxy encounter rather than the low-speed encounters we had considered.

While this did not absolutely rule out the hypothesis that VIRGOHI 21 was a dark galaxy, it presented a less exotic alternative - and as a general rule (known as Occam's Razor, after the medieval monk William of Ockham), the least exotic idea is considered the most likely explanation. But the simulations were far from being a great match to the observations, and further observations (unpublished) of the northern end of the stream showed it continuing at the same velocity as VIRGOHI 21 - more consistent with it being a leading arm than with the simulations. Could these simulations really explain VIRGOHI 21?

You're probably wondering why we didn't publish the new observations of the northern end of the stream, and point out the other inconsistencies between the simulations and the data. There are two reasons for this. The first is that while the simulations may not have been a great match to the data, they established the principle that it was possible to get long streams of neutral hydrogen via fast interactions. Once the stream was drawn out, there were many different forces acting within the cluster environment that could have altered its shape in myriad ways, so no simulation could be expected to provide an exact match.

The second reason is more complex, and is (to my mind) the most conclusive evidence against VIRGOHI 21. To understand this, it is necessary to go back to earlier in the story and look at predictions we made around the time the Arecibo surveys were starting up.

When we announced the Westerbork results, Jon Davies said in our press release that “We’re going to be searching for more Dark Galaxies with the new ALFA instrument at Arecibo Observatory. We hope to find many more over the next few years – this is a very exciting time!”

This wasn't just idle speculation - a year earlier we had published a paper, "The existence and detection of optically dark galaxies by 21-cm surveys", where we had argued that if VIRGOHI were a dark galaxy it implied the existence of many more dark galaxies. If these existed, they should be discovered by the next generation of neutral hydrogen surveys then getting started at Arecibo - and in large numbers. These dark galaxies would make up almost a quarter of the sources found in the deep AGES survey, while over a thousand would be seen by the shallower but much larger ALFALFA survey.

This meant that the final test of whether or not VIRGOHI 21 was likely to be a dark galaxy was whether other dark galaxies could be found. If there turned out to be a large population of dark galaxies then VIRGOHI 21 would be vindicated, but if no other examples were found it would imply that VIRGOHI 21 was nothing more than tidal debris.

So, what happened? To date, AGES has found very few confirmed HI sources without optical counterparts, and none like VIRGOHI 21. ALFALFA has found around 50 candidates - far fewer than required, and these are (as I understand it) potential mini-halos in the Local Group rather than objects similar to VIRGOHI 21. The only conclusion that can be drawn is that the predicted population of dark galaxies doesn't exist - which means, whatever the evidence for it individually might be, VIRGOHI 21 is highly unlikely to truly be a dark galaxy.

So - VIRGOHI 21 is almost certainly not a dark galaxy. Less exotic explanations have been given for its existence, and more recent surveys have failed to discover any similar dark galaxies. Does this mean we were wrong to publish what we did? No - the less exotic explanation was only discovered in response to our papers, and the surveys that should have uncovered more dark galaxies happened after our discovery. By publishing, we advanced science, even if those advances ended up showing that we were wrong!