No stacks please, we're simulationists

Back in 2017 there was a Nature paper claiming to have detected declining rotation curves in galaxies at high redshift. This would mean that galaxies in the distant early Universe have significantly less dark matter that contemporary nearby galaxies, whose flat rotation curves are one of the principle signatures of dark matter in the first place.

I was rather skeptical of this. None of their individual measurements looked the least bit convincing to me, with the fitted curves highly dependent on single data points : the slightest error could have thrown them off (and some curves just don't go through the points at all). True, the stacked curve was much more convincing, but any systematic error in estimating the rotation velocity will only compound this error rather than averaging it out.

On the other hand none of this actually would be evidence against dark matter. As with the now-plethora of Ultra Diffuse Galaxies and the like (in the nearby Universe, with relatively precise, sensitive data) which seem to have a significant and sometimes total deficit of dark matter, all this really indicates is that dark and visible matter can be separated. This is very much harder to do with modified gravity theories. If the rotation curve arises only as a result of baryonic matter, then if you have two systems in which the baryons have similar distributions (and are suitably isolated), then they should always have similar rotation curves. The only reasonable way in which they can differ is if one has dark matter and the other doesn't.

The real question is whether, according to cosmological theory, we expect galaxies in the early Universe to be less dark matter-dominated than today's. Ethan Siegel simply says "yes", that these declining rotation curves are indeed expected in standard models of galaxy formation.

The author's of today's paper, however, say No. And this is certainly more intuitive. If dark matter is mass-dominant, then it seems odd that it would actually become more important over time. Surely it should be gas falling into dark matter haloes that describes the process of galaxy assembly, and if that's the case – if dark matter makes up the bulk of the mass of galaxies from the word go – then they should always have similar (though not necessarily identical) rotation curves to those of the present day.

Now I should mention that the first author was my Master's project supervisor. You can read about this in some detail here, but in brief, we ran a galaxy formation simulation without dark matter and found that it just didn't work (see also my latter efforts failures to build a stable disc without dark matter). Anyway, the arguments that seemed quite compelling to younger me no longer have the same appeal; I'm delighted to see that others are still pursuing this line of inquiry and I wish them well, but I don't see this as likely to be anything more than a dead end. An interesting one to be sure, but I'm not convinced there's light at the end of the tunnel, so to speak.

What they do is run a series of simulations of the monolithic collapse scenario of galaxy formation. This is far simpler than the standard paradigm of hierarchical merging, in which galaxies assemble from a multitude of mergers in the early Universe (which can be a surprisingly efficient process, but then the early Universe was a lot smaller than the modern one). They consider different initial geometries, in which the initial dark matter halo is the same size, smaller, differently-structured, or spatially separated from its associated gas cloud.

They get the same result in all cases. The collapsing monolith experiences violent relaxation which essentially wipes out its initial conditions : the collapse proceeds at ever-increasing speed, the gas shocks and triggers star formation, and the galaxy forms during the re-expansion phase. This means that it doesn't really matter how you start, you get the same thing regardless.

And in this scenario you always get a galaxy with a flat rotation curve. The only way they could get a declining curve is to take out the dark matter altogether*.

* I'm intrigued that this is much more successful than my attempts at the same scenario, which gave something completely unphysical. I'll have to try and follow up on that.

This, as they've argued before, suggests that the initial conditions are irrelevant. But as I understand it, the hierarchical merging scenario is just more radically different than they give it credit for. There's just no reason in modern cosmology to assume the presence of any sort of collapsing monoliths, certainly at the very least not at all as the norm for galaxy formation in the early Universe. I don't think you even get much or any in the way of violent relaxation in this scenario, but rather a continuous assembly of tiny, already stabl-ish- proto-galaxies. So I really don't think it's fair to compare the failure of reproducing declining rotation curves in a monolith collapse scenario with the expectations from standard cosmology; this is comparing apples with oranges.

I also wonder just how similar these objects are to the observations. They don't compare the gas masses at all. Their gas radii are in some gases huge at 70 kpc : not outlandish by any means, but definitely on the larger side. They don't give this comparison to the observations, preferring to normalise their results to more accurately compare the shape rather than the size of the rotation curves.

This isn't a mistake in and of itself. But as in the press release that accompanied the original claim, there are a multitude of different factors in play. The galaxies detected might not actually be the progenitors of modern-day spirals but rather more spheroidal systems, which are anyway known to have less dark matter. There may indeed be significantly less dark matter in early galaxies overall, and those earlier systems might be more dominated by turbulence than rotation – which significantly affects the interpretation of the rotation curve.

This is not to say that there aren't questions to answer. The authors of this study point to other findings of more typical simulations saying that the turbulence isn't enough to account for all this, but they also note other observations of individual galaxies showing flat rotation curves (though they dispute this result) at earlier epochs. And they note that some galaxies might form in dark halos and others not, which seems very likely given all the hoo-hah about UDGs.

All this is very reasonable. The whole thing is a mess with many different variables to juggle. It just seems to me that we're a very long way indeed, further than ever in fact, for needing to invoke other physics like magnetic fields (as the authors do) to explain modern flat rotation curves. It seems far more likely that a combination of different effects are likely to explain early declining rotation curves than it is that they undermine the whole paradigm, which is otherwise supported by a vast array different and independent considerations, both in theory and observation, on a whole set of enormously different scales. 

Do we fully understand galaxy assembly ? Absolutely not. But it's way to drastic to point to a complex result like this and say it calls the whole basis of modern theory into question, and it's just not fair to compare the results of a monolithic collapse scenario with the radically different processes postulated by mainstream cosmology.

Theoretically dark

How do objects which don't form stars remain so dark ? That's a question I've often asked, especially of dark galaxy candidates : objects which have gas that looks like it's rotating (implying a dark matter halo to keep it bound together) but lacking any detectable stars. The big problems with such objects is that it's very hard to know if seeing is believing, if they just look like dark galaxies or were instead formed by different processes. Maybe such objects are just bits of gas ripped off perfectly normal galaxies. I've covered this in detail umpteen times before, with the basic conclusion being "some are, some aren't".

In the 2000's there was a period when a few different groups ran numerical simulations looking at whether specific candidates could be explained in this way. This has largely died off, so I was very intrigued by this paper which examines dark galaxies in the context of the latest and greatest numerical simulations. These are completely different beasts from the n-body simulations run on the desktop machines of 20 years ago : instead of a few thousand SPH gas particles, now they have billions or more particles and included all kinds of fancy gas physics that would have had us all foaming at the mouth in an ecstasy of delirium back in the day.

It certainly starts in a promising way, reviewing the major candidate objects and studies (and yes, they cite me, so thanks for that) as well as some other more recent work I wasn't aware of. So I've got a couple of other references I should check up on, which is good. But I have to say that after that it's all rather more theoretical than what I was hoping for. Not that it's difficult or unimportant, but that it never makes any comparison between theory and observation. It deals with the dark galaxy candidates in the simulations very much on the terms of the simulation alone, making little or no comparisons with the observational candidates.

In some ways this is quite novel, at least to me. Normally I look at the missing satellite problem from the perspective of the galaxies, because those are what we actually observe. But the problem itself is all about how simulations predict too many dark matter halos that never light up, so examining those halos as interesting objects in their own right is a good idea.

What they find isn't terribly surprising though. The vast majority of the halos in the simulation do indeed remain dark, for what seems to be due to a combination of factors more than any one in particular. And they form a continuous sequence from the truly starless to the merely very dark to the brightest and most luminous objects of all; dark galaxies are not special, but normal. Indeed, perhaps we should instead be asking instead not what keeps some halos dark but the exact opposite : what allows such extreme levels of star formation in the apparently "normal" galaxies ! For comparison, in their simulation they identify 5.6 million halos, of which 5.5 million are completely starless, 47,000 are dim but not totally dark, and the rest – a mere 100,000 or so – are luminous.

If we stick with the standard question, "what keeps them dark ?", though, then the answer seems to be : isolation, spin, and mass. Isolation prevents them from experiencing as many mergers as the brighter galaxies, which compress the gas and trigger star formation. Isolated objects avoid this. Spin keeps the gas more extended and its density lower, thus reducing star formation. And mass prevents much gas from getting into the halo in the first place, again keeping density low. While some dark galaxies do form stars briefly early on and then lose their stellar population, it seems that most just never form any at all. 

There's an additional effect from mass. Being small means that objects are more vulnerable to the effects of reionisation : when the first, highly energetic stars light up, they ionise all the gas in the smallest halos* and drive it out, and being so small they don't have the gravitational strength to recapture it. 

* These population III stars are thought to have been true behemoths, much larger and more energetic than any stars around today. So these wouldn't necessarily have had to form inside the halo that would later become a dark galaxy, they just had to be in some reasonably-nearby larger galaxy.

And that's really all there is to it. They cover this in great quantitative detail, much of it having long been examined before but here all at once and in some depth. But how does one go about verifying this ? How many such dark halos should have enough gas to be observable with current HI surveys ? How do the line widths of the candidates compare with the theory – how many should we expect to see according to the model ? How does this quantitatively address the missing satellite problem ? What testable predictions does it make ?

Frustratingly, none of this is mentioned. It's great to see dark galaxies being used as a mainstream term but it feels like a cliffhanger ending, stopping at the point things get interesting. And I seem to recall other people having problems with making the reionisation ("squelching") solution fit the observational data, so more comparisons to earlier works would have been nice. 

Still, the idea of dark galaxies, being a once openly-derided solution to a major problem in cosmology, now seems to have transformed into an inescapable inevitability, not a problem but simply reality. Specific candidates, I suspect, will always remain problematic, but the notion in principle now appears to be greeted nor with mere tolerance but actually embraced : yes, these halos do exist, it's just that we can't see them directly. So the wheel turns.

The faintest galaxy is getting even fainter

Today's paper is about a galaxy so faint that after reading the discovery paper back in 2018, I must have immediately forgotten all about it. Which was a mistake, because it's in Leo. Since I did a whole paper on optically dark gas clouds and stuff in Leo but didn't cite these authors, I now feel a little less annoyed that they don't cite me either.

Anyway, this little galaxy's claim to fame is being bloody dim. It is in fact, they say, the faintest object every detected by its optical emission. It's embedded right in the heart of the Leo I group, home to the extraordinary Leo Ring, which is just a little to the north of this object. In fact, the galaxy is in a bridge of HI connecting the Ring to the giant spiral galaxy M96, whose own HI emission is rather distorted.

How does such an incredibly faint, diffuse object form ? They suggest two scenarios. It could be a really extreme but basically "normal" galaxy, having managed to maintain an incredibly low star formation rate over the whole lifetime of the Universe. This would be really interesting because it would then be unclear how it ever formed any stars at all, with models not predicting stuff like this : it seems to be just too diffuse to do anything (this might be a fun connection to other Ultra Diffuse Galaxies found elsewhere). Or, perhaps less excitingly, it could be a tidal dwarf galaxy, having only just recently formed from the gas stripped away from M96.

If it was the first scenario, then it should be expected to have an old stellar population. So here they present Hubble observations indicating pretty unequivocally that it doesn't. They can detect the occasional older star but nothing above background contamination levels, with the stars in this object clearly dominated by younger ones. It all suggests a brief starburst took place about 300 Myr ago and then stopped, with pretty much nothing happening today at all based on the Halpha emission. All their measurements emphatically support this : the distribution of stars, their metallicity, modelling of these parameters by different techniques, everything. Circumstantially, the distortion of the gas in M96 also supports this interpretation (though we still have no idea how the Leo Ring itself, which is connected to M96, actually formed).

So this is pretty cut and dried. The tidal dwarf model means that unfortunately this doesn't really tell us much about galaxies more generally, since this would be enormously atypical of the galaxy population in general. And in fact they describe this galaxy as "failing", being unable to sustain further star formation and likely to be tidally disrupted. What seems to have happened was that the initial compression of the tidal field triggered a brief burst of star formation, but as the whole larger system expands, this has already stopped.

This all raises two questions for me. What does this imply for the optically dark clouds in the same region ? Remember one such "cloud" is actually optically bright, and not especially diffuse by any standards, either optically or in terms of gas. It's an open question whether that object is in any way related to the other, truly dark clouds or just a coincidence that it's so close to them and so similar in terms of its gas content. Could it be a similar case to this galaxy but far more extreme ? I dunno.

The second question is, why isn't there a third option ? Why can't it be a hybrid of the two ? Rather than a normal but extremely diffuse galaxy "recovering" from a tidal encounter, could it not instead be a once-dark galaxy that has only just lit up thanks to the tidal encounter ? I can't think of anything here that would rule this out, but nor can I think of a good way to test this. Still, maybe things aren't quite as clear-cut as they at first appear.

El Gordo plays ping-pong

We've encountered "El Gordo" (The Fat One) a couple of times before. First there was a claim that this merging cluster is just too big, too early in the Universe, and colliding at a speed too high to be compatible with Standard Model predictions. I noted that such claims do seem legitimate, but I wouldn't jump on the bandwagon just yet : the chance of finding such a cluster seemed to be very non-linearly dependent on the parameters, so I'd bet observational errors could have quite a part to play here.

Next there was a rebuttal which said that everything was fine. The mass was actually only about half the earlier estimates, and the infall velocity way smaller. But while I'm strongly in favour of the Standard Model and naturally suspicious of these claims that it's been debunked, I have to say I didn't like this particular paper. Their mass estimates seem fine so far as I can tell, but I found the justification for their much lower collision velocity very badly expressed.

Along comes a counter-rebuttal (hence the ping-pong) that, oh joy, does the same thing but for the opposite side of the argument.

The authors of this latest study don't seem to have any beef with the revised lower mass estimate. Instead their attention turns to the infall/collision velocity. They maintain there's also still a problem with the more famous Bullet Cluster, but the studies I read a while back found that there isn't, so I'm going to ignore that aspect.

They start with quite a nice overview of previous studies but immediately fail to learn any lessons from them. Noting that different authors have come up with radically different values and interpretations, they seize on the latest measurements as being of unimpeachable accuracy and precision. Inclination angles have varied from 30 to 75 degrees or so, infall velocity estimates have varied from 1200 to 2500 km/s, the current stage of the merging has been interpreted differently... all this to me suggests that there's really quite a wide margin for error here, and we just don't have good enough observational data or numerical simulations to constrain anything very much,

I also dislike their whole interpretation of the low probabilities of finding such objects in the Standard Model. It's a one-in-ten-billion, they say. This is a big red flag by itself. It doesn't make any sense to me that the Standard Model could get things so fabulously right in such a plethora of circumstances and then fail so utterly miserably in others, where the forces at work are supposedly the same. Of course, to give them their due, they would say that the Standard Model fails miserably elsewhere too, but they're wrong about that.

On a more pragmatic level, I wish they'd define the velocity terms more clearly. What exactly is meant by the infall, peculiar, and observed velocity ? Oh I'm sure they're simple enough (I can certainly make an educated guess what they mean), but without having them spelled out and rigorously defined, this becomes just as confusing as the previous paper.

And where this really becomes critical is their underlying methodology. The previous authors, they say, deliberately disregard high infall velocities because that contradicts the Standard Model. They say this means we can't use that analysis to say how likely it is that such a feature arises in the SM, but I think this is daft. If we can show that such a scenario works, that this agrees with observations, then the circular nature of the argument doesn't matter : "we looked for SM-compatible solutions and found one" is a perfectly valid approach, and while it might not tell you about probabilities, demonstrating compatibility is much more important.

And then they do the exact same bloody thing but in reverse, insisting that they must search simulations only for objects with infall velocities which are incompatible with the SM ! This is rank hypocrisy and exasperatingly silly.

They make similar daft and blunt claims about other previous analyses too. They could have just said "we improved the analyses" rather than stating that the others were outright wrong. Saying there's no reason to assume the cluster components are gravitationally bound in no way implies that they're not. And sure, it could be that LCDM is incorrect and it could be that this would result in a high infall velocity, but what exactly is the reason to assume with such certainty that this must be the case ? When is a factor of a mere 1.5 grounds for an unsolvable discrepancy ?

It's all just weird and I don't like it at all. If you're going to say that some infall velocities must be excluded, then say why. Saying that you exclude them just because they're compatible with the SM makes not a lick of sense. That's committing exactly the sin they were trying to avoid. I liked their earlier paper much more than this. As it stands, I'm now far less convinced that El Gordo is the CDM-killer they think it is.

You should have gone before we came out

Where do galaxies lose their gas ? Galaxies in clusters have been shown to have much less gas than in the general field. In Virgo, for instance, on average they have ~50% or less than comparable field galaxies of the same morphology and brightness. 

Now we know there are lots of processes at work inside clusters themselves than can do this, especially ram pressure stripping. That's almost certainly responsible for the majority of the gas lost. But there are many hints that galaxies elsewhere can lose gas too, not usually as much as in clusters, but enough to be significant. This is important because most galaxies don't live in clusters, so if we really want to understand galaxy evolution, we should probably stop spending so much time on the sexiest 1%* of the population and look at all the others from time to time.

* Well, it might be a few percent, but not more than this.

This is another paper I wouldn't normally read because reprocessing existing data tends not to accomplish much. But in this case I think they're on to something... ironically by looking at clusters. Though I have to say, their sample definition seems convoluted in the extreme and they gave such a detailed breakdown of how they divided everything I wanted to slap them and shout JUST TELL ME THE DAMN NUMBERS ! Which they eventually do, and it's a few thousand per object type.

Key to this is that they can distinguish between different sorts of galaxies in clusters. I was racking my brains because this seemed curiously familiar. In fact this idea was mentioned a couple of years ago in a conference, but disappointingly the presenter isn't on the author list or cited. It may, of course, be a completely independent discovery.

Anyway, most galaxies in the field are in small groups, often very small : two or three members, though sometimes more. So this means that clusters tend to assemble by absorbing whole groups of galaxies rather than individual ones. In some cases it's possible to identify subgroups within a cluster, which are therefore likely recent arrivals. Individual galaxies within clusters, not part of any subgroup, could be individual field galaxies falling in for the first time, but they're more likely to be older arrivals whose original constituent group has been broken apart by the chaos of the cluster. 

By analogy, nightclubs. It takes a while for groups of people to get broken up by the general throng, so if you see a group of people all together, chances are they just arrived. And equally, not many people go to nightclubs by themselves, so solitary people looking lost and confused have probably become unfortunately separated from the group that originally dragged them in there.

But... galaxies. What the authors do here is very simple. Using samples that identify these different galaxy types, they plot (see their figure 1) the fraction of star-forming galaxies as a function of cluster-centric distance. They find that for solitary galaxies in clusters, the star-forming fraction steadily increases as you go further from the cluster... right up to the level of the general field. There's no distinct break in star formation activity for individual galaxies as they enter the cluster, it's just a smooth curve.

In contrast, galaxies which are still in sub-groups within clusters also increase their fraction of star-forming members, but they reach a plateau. At high enough distances, the star-forming fraction never increases for these galaxies, whereas for the solitary ones it just keeps rising.

In other words, the gas-rich, late-type (spirals and irregulars) individual galaxies just continue losing gas as they get ever-closer to the cluster centre. But exactly the same type of galaxies which are in groups never had as much gas to begin with. They've already lost some of their gas. Not as much as they eventually lose in the cluster, but still a detectable difference compared to individual galaxies. Ergo, pre-processing definitely happens.

I think this is a very nice confirmation of something already strongly suspected. The main accomplishment here is using a substantial sample to increase the cluster-centric distance well beyond what was previously attempted, which explains why it wasn't seen before. You really need to go to very large distances indeed to see this, but when you do, it's astonishingly clear. Honestly it's rare and gratifying to see a result in astronomy which is so clear-cut as this. Hooray !

One small caveat : you might be wondering, well, shouldn't the individual galaxies still show a plateau if they too tended to be in groups to begin with ? Probably not. Deep within the cluster all galaxies will be dominated by cluster-specific processes. But at the distances where there's a difference between individual and group galaxies, it's probable that the solitary objects were never in groups : they're just too far from the general melee of the cluster to have been much affected by it. So there is a change in the nature of individual galaxies as you descend into the cluster, from being dominated by the habitually-solitary to those who were indeed once in a group. But both of these end up being equally affected by the cluster's ram pressure, hence the smooth, continuous change in star-formation properties.

Do legumes give galaxies gas ?

This paper is about "green pea" galaxies, which are so-called because they're small, highly concentrated and even look green because of their strong spectral line features. Green is a very rare colour in extragalactic astronomy, and indeed for stars in general. The blackbody curve over which they emit means that they emit so much light across the whole spectrum that green is always washed out by the blue and the red. Only when you get something that's not a blackbody, something emitting over a very narrow wavelength range, do you get anything green. And that's rare.

"Green peas" tend to be more distant, but there are also closer "blueberries", which you'll have guessed are much the same apart from the colour. Whether they have similar spectral line emission to green peas is not clear.

Anyway, the authors here try and figure out how the HI (atomic hydrogen) gas content of the green pea galaxies varies with their stellar content. A perfectly sensible goal given how strange these objects are; perhaps this could shed light on star formation in extreme objects. But I have to say I don't like it very much.

Like some other papers I've been reading lately, this one contains no new observations but uses entirely archival data. But unfortunately this one is more typical of its class, not actually finding anything new or at most a marginal, incremental difference. They have a sample of 19 HI detections (and 21 upper limits, which were observed but not detected), though they have to exclude a couple because of issues with the optical and/or UV data. Based on the details they go into, they seem to have re-processed the HI data, though it isn't at all clear why – was there some issue with the earlier analyses ? If so, they don't mention it.

Figure 2 is by far the most interesting. This plots the now-classic "main sequence" of galaxies showing how their star formation rate scales with stellar mass, in a nice, tight correlation. Green pea galaxies also seem to correlate but with a totally different, much steeper relation : they have far higher star formation rates than their stellar masses predict. This holds true for both the HI detections and non-detections. Unfortunately the upper limits of the non-detections are all over the place, so it's not possible to say if there's any real difference between those with and those without gas. And the main trend is not a new result.

What they instead concentrate on is finding other relations to the optical data. They say their sample is offset from a previous relation between the MHI/M* relation as a function of NUV (near UV) - r magnitude (basically a colour measurement*), but... well it might be, but to me that earlier relation, for normal galaxies, itself looks like a dodgy fit. So I don't think too much can be said here.

* One minor but quite interesting point they make is that using this UV-optical is better than using two optical wavebands for colour, since the green pea star light is much more UV-dominated than in normal galaxies.

One thing they do show which looks convincing, but I'm not sure if it's a new relation, is that green peas have excessively high gas fractions as a function of just about any parameter. New or not that's nice, but even here their plot could easily have been so much clearer. Then they try and plot gas fraction offset (measured – expected) as a function of various parameters, say there's a trend, but as far as I can tell there just isn't. This is one of those cases where if someone shows you a weak trend and you say, "I've seen worse claims"... this is one of those worse claims.

Then they try and find what sort of scaling relation does the best job of predicting the HI mass of green peas. As far as I can tell, they seem to have taken the standard practise of throwing everything against a wall and seeing what sticks a little too far. They come up with hideous relations involving different colours, stellar masses, star formation rates and surface brightness levels that to me has no obvious physical significance at all. Frustratingly, they don't discuss this. 

And that I do find strange and annoying. It's reminiscent of p-hacking, where if you search for correlations using complicated enough relations and large enough data sets, you're bound to find something. Was there any prior reason to suspect this torturous relation had some physical significance ? If so then it's interesting ! If not, if, as I suspect, it's just an empirical fit, then it's just a statistical artifact. I'm not saying that's the case, only that they needed to be a lot clearer about what this is supposed to mean, physically, before I take any further interest in it.

It's gotta come from somewhere

One thing that's never seemed terribly puzzling to me is where galaxies get their gas. Estimates of the star formation histories show that with their current gas content, galaxies should typically exhaust their gas within a gigayear or so. A lot of people find this suspiciously fast, that to maintain their currently constant levels of star formation is impossible unless the galaxies are being re-fuelled from somewhere.

Personally I've always found that one gigayear is not fast enough for a need to invoke refuelling (or accretion as it's usually known). It just means that galaxies had more gas in the past and will run out comparatively soon, but in a time equivalent to about one-tenth the age of the Universe isn't enough to set alarm bells ringing for me. Sure, if it was the next few or tens or even hundreds of millions years, then I might be concerned. Then I might say, "hang on, isn't it a weird coincidence that we've arrived on the scene just as galaxies are about to stop forming stars forever ? Isn't it more likely they're being replenished from somewhere ?"

On the other hand, the apparent constant rate of star formation does seem more legitimately odd. If galaxies are truly running out, naively you'd expect to see that reflected in their star formation histories. So people have postulated that galaxies are accreting gas from the field somehow, either from "hot mode" where gas cools very slowly and omnidirectionally, or in "cold mode" where it condenses into cooler, more distinct streams which funnel themselves into the galaxies. Claims to have detected the latter are always controversial because it's very hard to say that a stream of gas isn't just gas that the galaxy is losing by a host of much more well-understood processes.

Today's paper attempts to address this mild puzzlement. They use a sample of galaxies which is as homogenous as they can get and apply some reasonable scaling relation where necessary to calculate the change in gas. For instance, they assume that the bulk of the normal, "main sequence" galaxies today were already on the main sequence 4 Gyr ago. That is, they formed stars at a predictable rate given their total mass of visible matter. This is not at all an unreasonable assumption : 4 Gyr is enough to expect some evolutionary changes but nothing dramatic, and while of course plenty of individual galaxies might have experienced the odd burst or sudden cessation of star formation, there's no reason to think these would be statistically significant. 

And kudos to the authors for clearly acknowledging their assumptions and how complex real star formation activity can be. In public talks I sometimes go on a breathless monologue describing the various process at work and how they relate to each other; they do much the same here, except that they clearly spell out how this is likely to affect star formation – rather than my own take-home message which is only that it's bloody complicated. 

The really big assumption, the most difficult point to make reasonable inferences about, is how much the HI gas has changed. A handful of stacked observations have now managed to detect this atomic gas out to these vast distances, but these combine data from hundreds of galaxies. We currently have no real idea how it's changed in individual galaxies on these timescales; in contrast, the molecular gas can be observed directly and is much better understood. 

Add to that that the relation between atomic gas and star formation appears to be subtle. It's popularly described as the galaxy's fuel tank, the reservoir from which star formation ultimately occurs. Which is quite appropriate, since understanding how much gas is in the tank and how fast a car is going is not easy ! The actual gas in the engine itself, the stuff that's exploding in the pistons, is thought to be molecular, but there are strong hints that HI is directly involved as well at least in some cases. Most HI-rich galaxies are blue and star-forming, but some have loads of gas but hardly any stars at all or only old red ones, so it's not at all a straightforward connection.

Using their various scaling relations as best they can, their conclusion is at least an interesting possibility. They say that what seems to be happening is that galaxies are both losing and gaining gas : overall, they're running out of the molecular gas (the stuff in the engine itself actually doing the business of star formation) but actually gaining atomic gas (the stuff in the tank). This is a small change for dwarf galaxies but really quite large (70 % !) for massive ones, even to the point where the galaxies have grown significantly in overall baryonic mass as a result of this. Galaxies then, are still assembling even today, not just from mergers but from the condensation of the thinnest gas in the intergalactic medium.

What this means is that something is changing star formation efficiency. Somehow galaxies in the early universe were able to efficiently convert all their gas into stars, whereas today something is preventing the HI from cooling into molecular gas which can form stars.

Why could this be ? Frustratingly they remain silent about this, which is annoying because this is such an obvious question you can't not ask it. Especially since modern galaxies are much more metal-rich, which enables very much faster cooling and condensation of the gas : if anything they should be more efficient at forming stars, not less. On the other hand there are far more stars around today (though less energetic), so perhaps stellar feedback is to blame.

Well, I dunno. Fair play to 'em for being clear about all their many assumptions, but where we go next with this is anyone's guess. It needs a lot more independent studies before anything else can be said about this. 

Dotting the i's and crossing out the dark matter

This paper revisits one of our old friends, those galaxies without dark matter.

When last I checked in on this, it seemed to be settled that indeed they do lack dark matter, and they can be explained by conventional (though rare) interactions which can strip the dark matter but leave behind a surviving stellar core. This is probably still the case. The whole distance debacle appears to have been decisively settled such that the low velocity dispersions of the objects are indeed consistent with little or no dark matter. The authors note in the introduction, however, that a few diehards maintain that maybe we're just seeing them close to face-on, which would hide the signatures of rotation.

Dedicated readers will recall that I myself thought this quite plausible for some other, similar objects, until a recent paper finally convinced me that this just isn't tenable. This is why it's important to check every hypotheses as carefully as possible. The only thing that ever really settles the arguments is better data.

So what the present authors have done is gone and get measurements of how fast the stars are moving in the notorious NGC 1052-DF4. Already the globular cluster data is clear that it can't possibly have dark matter, with a velocity dispersion of just 4 (!) km/s. But direct measurements of the stars in the galaxy itself would be much more decisive, because nobody's ever really going to be happy with data from just seven globular clusters.

And the measurements, if sufficiently precise, can help beyond settling the major issue. As they say, different formation scenarios (such as tidal stripping versus collisions) are expected to result in different amounts of dark matter remaining, though always of course on the low side. Determining just how low this is requires extremely precise data, the kind you can only get from the stars themselves.

So that's what they go and do. The have 14+ hours on the Keck telescope and find the measured velocity dispersion, though larger than from the globular clusters at 9 km/s, is still entirely consistent with the galaxy having no dark matter whatsoever, and inclination angle effects can be neglected. Making this measurement is a much harder task than in regular galaxies, because here the dispersion is so low that even motions of individual stars need to be properly accounted for. After various corrections, their final estimate of the true velocity dispersion is just 6 km/s. The traditional NFW profile for the dark matter for an object like this would give a mass about three orders of magnitude greater than what they observe.

(I confess to being a little caught-out here. The globular clusters are found further away than the stars but their velocity dispersion is lower...? So used to declining rotation curves am I that this seemed really weird ! Then I remembered this is exactly what's supposed to happen according to Kepler)

They also give a much better explanation as to why the globular cluster population of these objects is interesting in itself, compared to other papers which I find never get the point across. For normal galaxies there's a nice simple linear relation between total dark matter mass and number of globular clusters. These objects, according to that relation, are consistent with much more massive dark matter halos. That is, they have far more globular clusters than one would expect based on their small/zero mass of dark matter. They don't fit the standard models of galaxy formation at all. 

In fact they're outliers in at least three different ways : they have less dark matter than expected given their stellar mass; more globular clusters than expected for their stellar mass; and more massive individual globular clusters than is typical. And their globular clusters are, they say, "remarkably consistent in colour".

They don't here speculate much as to what all this means for the formation scenario, and in this case, given the controversy that has engulfed these objects, I can't say I blame them. They do however note that there is some tension with cored dark matter profiles as well as the standard NFW ones, though it's a rather weak tension. More interestingly, they say that the velocity dispersion is "clearly inconsistent with MOND", which predicts > 12 km/s. This is fun because previously claims of inconsistency were shown to be premature because they hadn't accounted for the external field effect, a MONDian effect whereby nearby galaxies can change each other's dynamics in a way that doesn't happen in standard gravitational models.

We shall see where that one goes in due course, I'm sure. Still there are many questions about these objects. Since there are two in the same group, the presumption is they must have formed the same way. And somehow they have to have survived in the group without being destroyed by tidal forces, which is counter-intuitive for large, low-density objects. So if the major issue is settled, it hardly feels like we've heard the last of these yet.

Is it a group ? Is it a cluster ? No, it's a supergroup !

This paper caught my eye for the wrong reasons, but it turned out to be interesting all the same.

It's time for another look at our old friends, Ultra Diffuse Galaxies : those faint, surprisingly large smudges which have been confusing us all for some years now. I've covered their dynamical masses umpteen times before, but their gas content is interesting not just in understanding their dark matter content, but for their star formation too. So when this study said it was using an HI survey to examine UDGs, I was hoping for more than one measly detection.

They start with a nice little overview of possible ways to form UDGs. They might be dwarfs which were spinning unusually fast, so distributing their gas over such a large area that its density was too low for much star formation. Or it might have been early star formation that drove the gas out to larger distances, with the effect being the same; likewise tidal encounters could do something similar. More dramatically, they might be as massive as Milky Way-sized galaxies, with something happening to quench their star formation in a process yet to be understood.

The authors are using the WALLABY survey on one of the SKA pathfinder telescopes, which gives decent resolution and good sensitivity over a very wide field of view. Their target is the "Eridanus Supergroup", which is fascinating in itself. Only a few such supergroups are known, and they're thought to be groups in the process of merging to form a full-on cluster. 

And they really do blur the boundaries between groups and clusters. One individual group in Eridanus has hot X-ray detected gas despite its small number of constituent galaxies. I'm going to have to update my introductory talks to include stuff like this – the presence of X-ray gas is a common way to distinguish groups from clusters, besides the more-obvious parameter of sheer number of galaxies. A small group which has this hot gas is not at all typical. Though it would have been nice of the authors to give a table describing the properties of the various Eridanus sub-groups; the maps that they show are very crude and not useful.

The X-ray component seems to be quite substantial, since galaxies here have measurable levels of HI deficiency. In clusters this would usually be interpreted as a classic signature of ram-pressure stripping. Yet with only a few galaxies, rather than a few tens or few hundreds as in most clusters, it's anyone's guess where this hot gas actually came from in the first place. They also note that there are two "enormous HI clouds without optical counterparts" in Eridanus, although that's not quite how I would describe them.

Anyway, they make a search for UDGs using the usual criteria and find 78 candidates in the WALLABY HI survey region. Since they don't have redshifts, most of these are probably background misidentifications but that's okay. But because this catalogue comes from something designed to search at greater distances, they apply an additional size constraint which leaves them with just 6 candidates.

Then they do something a bit strange : they search the HI cube using an algorithm. I mean that's fine, perfectly fine... but... it's not ideal in this case. Far better, I would think, would be to extract individual spectra of each UDG and see if there's any hint of a detection. After all, in this case the search is specifically for UDGs, so there's no need to insist the HI catalogue be homogenous – this is unnecessarily restrictive. Searching cubes blindly is all well and good but you have a much better chance of detecting the faintest stuff if you already know where to look. You want a blind search if you're specifically interested in which objects are gassy, not if you want to know how much gas each object has.

This all makes it a bit odd to call this a WALLABY paper since hardly any of the rest is concerned with the HI at all (they don't describe anything about their resulting HI catalogue, leaving me wondering if they only detect the UDG but that would be remarkable in itself). I mean they do try, but I think all the rest of the interesting stuff is in their discussion of optical relations, with the HI very much being an aside.

And there is interesting stuff here to be sure. They calculate the expected number of UDGs in each subgroup, which apparently follows a power law according to the group mass. They also show that (from other studies) the slope of this power law is invariant with environment, suggesting that environment doesn't matter much to the formation of UDGs. Which is very interesting considering that UDGs as a whole are a very diverse bunch, with some being blue and structured and some red and smooth (and a couple here, interestingly, which are blue and smooth).

While the numbers of predicted and candidate UDGs are consistent for two of the subgroups, for the one with X-ray gas there's a discrepancy : 17 predicted, none found. They suggest that in this presumably more mature group, most of the UDGs have already been disrupted or merged with the central galaxy. In that case though, I'd want to know where this power law comes from exactly, since most UDGs have so far been found in clusters : it seems a bit paradoxical to find the biggest disagreement in the group that's more like a cluster than the others ! 

And they also say this isn't a surprise anyway because of the number of candidate UDGs there is small, but this makes no sense to me at all. I cannot really get my head around it. The claim seems to be that they can predict the number using the mass of the group (an independent value), but that there's a discrepancy is because there's not as many found as they expected... huh ? The reason for the disagreement is that the numbers are different ? That feels like a truly weird tautologous statement. I may well have missed something. 

While they make a brave attempt to consider if tidal stripping or ram pressure was most likely responsible for the UDG formation, with six objects this is inevitably a forlorn hope. Though they do find that one of the candidates is at the end of a long stellar tail in a pair of interacting spiral galaxies. I'd have made a bigger deal out of this.

I can't escape the nagging feeling that this paper is doing all the right things in the wrong ways, or at least in the wrong context. A blind comparison of HI detections and UDG candidates is great if you have substantial samples of both. Examining the scaling relations of UDGs in different environments is useful when you have novel data sets to present. 

As it is, they have two interesting things : the detection of a single UDG in HI (showing that these are still very unusual) and a UDG at the end of a tail. Personally I would have concentrated heavily on the details of these two individual objects. Show me the HI spectrum ! Tell me more about how that near-tidal UDG compares to other UDGs ! Leave the scaling relations for when you have more things to relate : individual objects are still themselves worthy of study; not every study has to be about whole populations.

To find HI, first eat cookies

Just a quick one because this is well outside my field but I absolutely cannot ignore it.

This simply wonderful paper starts with the utterly brilliant name of "WTH! Wok the Hydrogen" and it continues in that vein for the whole paper. It's all about building a radio telescope for public outreach purposes out of old bits and bobs you can find lying around in your kitchen. Of course, the main feature is the wok itself, which serves as the dish. Is it a parabola ? Is it a sphere ? They're not quite sure, but suspect it's neither because woks aren't normally built with radio astronomers in mind.

In any case it doesn't matter. With a diameter of 60 cm, only a few times the 21 cm emission this thing can detect, it has a 24 degree beam. The photo of the telescope sitting in a plastic tub that you could find in any hardware shop is by far and away the best observing setup I've ever seen. With a beam that large, you really just have to aim it vaguely in the direction of the sky and it'll do its thing.

It's not just a wok and the plastic containers though. There's also the cookie box, used for shielding the electronics. Apparently this makes a big difference to the sensitivity, allowing them to observe from the bustling metropolis of Hong Kong and still get a detection despite the billion or so mobile phones and other radio nasties in the area. They say this shielding is what makes the difference compared to other amateur radio projects that have had to go to quieter sites for detections.

Incidentally the use of the wok specifically is not just because of its ready availability in Hong Kong, but because its shape (whatever exactly that might be) is better than a traditional satellite dish because it has a better focal length. And one further piece of household equipment they use is a microwave oven, which is a ready source of copper wire for the dipole antenna itself.

How well does it work ? Look, this thing is never going to detect a pulsar. But it can detect the HI 21 cm emission from the Milky Way in 10 minutes. Point it towards the galactic plane for 10 minutes and BAM, a detection results. Point it away for the same amount of time and BOOM the signal goes away. For public outreach that's brilliant. You can't really do any more than that; they say it's just not good enough for mapping the galaxy or measuring a rotation curve. But who the hell cares ? They've taken a bunch of old junk and turned it into a $150 radio telescope that can detect gas in the Milky Way. That's a missed opportunity for Scrapheap Challenge (Junkyard Wars for Americans) if ever there was one.

And the best bit ? Our institute director wants to build one. Next year's open day should be interesting indeed.