The Elephant in the Universe, page 25
However, it has been hard, if not impossible, to solve the missing-satellite problem in a satisfying way. The numbers just don’t match. Given the masses and luminosities of the observed satellites, ΛCDM predicts that there should be quite a few larger and more massive subhalos that definitely should have turned into conspicuous dwarf galaxies—a discrepancy known as the too-big-to-fail problem.
Dwarfs misbehave in other ways, too, further complicating the dark matter search. Consider the effort to estimate the dark matter contents of dwarf galaxies by studying the rotational velocities of the stars and gas clouds within them. Such measurements were pioneered in the 1980s by University of Arizona astronomer Marc Aaronson, who was killed in 1987 by a freak accident in the dome of the 4-meter Mayall Telescope at Kitt Peak National Observatory. Around the turn of the century, the project was carried out in more detail by John Kormendy and Ken Freeman, who collected and analyzed data, obtained by others, on a wide variety of galaxies.
What Kormendy and Freeman found was that dwarf galaxies contain a higher fraction of dark matter than big spiral galaxies do, and dwarf galaxies are also more densely laden with the stuff.5 So far, so good: this is in good agreement with the ΛCDM computer simulations. But these same simulations also produce subhalos with a very characteristic density profile: as you move toward the core, the dark matter density increases faster and faster, until it reaches a peak value at the very center. This density profile appears to be an inescapable result of hierarchical clustering, at least in the supercomputer simulations.6
The problem is that real dwarf galaxies do not show these prominent density cusps at their cores. The dark matter distribution, as derived from velocity observations, is always much flatter. This third mismatch between ΛCDM simulations and the real universe is called the core-cusp problem, or the cuspy halo problem. Again, there’s no easy explanation. Dwarf galaxies just don’t obey the rules. Or, put differently, our rules do not properly describe reality.
The Dragonfly Telephoto Array shed surprising new light on the relation between dwarf galaxies and dark matter. However, dark matter was not on Pieter van Dokkum’s and Bob Abraham’s minds in 2011, when they first discussed the possibility of using off-the-shelf photographic lenses to image extremely diffuse structures in the night sky—faint wisps of nebulosity, but also low-surface-brightness galaxies. As an avid nature photographer, van Dokkum had heard about a new professional 300-mm telephoto lens produced by Canon, featuring a nanotechnology-based coating to reduce light scattering. That sounded perfect for wildlife and sports photographers working with backlight. It also sounded perfect for low-contrast deep-sky photography.
“Off-the-shelf” doesn’t necessarily mean cheap. The lens sold for approximately $10,000. But hooking up several of these lenses would still be much cheaper than designing and building a special-purpose telescope. By aiming many lenses at the same part of the sky, each mounted on its own professional CCD camera, you could digitally add up the individual images to further improve contrast and sensitivity. Thus the idea for an astronomical telephoto array was born.
It didn’t take long to come up with an appropriate name for the project. A big array would look more or less similar to the compound eye of a dragonfly, as van Dokkum knew well enough: ever since he was a young boy in the Netherlands, he had taken thousands of macro images of the beautiful insects, and he was working on a book.7 Dragonfly it would be.
What started out with test images in van Dokkum’s New Haven basement and backyard, using just one lens, soon grew into a working prototype of three lenses on one mount. Hardly more than a year after the Toronto dinner, he and Abraham moved their equipment from the Mont-Mégantic Dark Sky Reserve in southern Quebec to the even darker New Mexico Skies Observatories, in the Lincoln National Forest east of Alamogordo, where semiprofessional astronomers from all over the country operate dozens of remotely controlled telescopes.8
Meanwhile van Dokkum and Abraham recruited students to join the Dragonfly project and to build hardware, develop software, process the data, and analyze the results. The array quickly grew from three lenses to eight, then ten, then twenty-four on one telescope mount—an impressive compound eye indeed. Before long, the team constructed a second dome with a second twenty-four-lens array. The forty-eight telephoto lenses have the same overall collecting area as a virtual 1-meter telescope, but the focal length is still a mere 40 centimeters, yielding an incredibly “fast” optical system with a focal ratio of 0.4—a system able to register small quantities of light in a short span of time—something that can never be achieved with a single lens or mirror.
Part of the Dragonfly Telephoto Array at New Mexico Skies Observatory.
Dragonfly may have started out as a hobby project, but it soon evolved into a novel, high-profile robotic observatory that uniquely focused on the much-neglected low-contrast and low-surface-brightness universe. Not surprising, then, that it started to yield exciting scientific results right away. Dragonfly images of the Coma Cluster of galaxies revealed forty-seven extremely faint smudges of light, the vast majority of which had never been observed before. In their 2015 discovery paper in The Astrophysical Journal Letters, van Dokkum, Abraham, and their colleagues called the smudges ultra-diffuse galaxies, or UDGs.9 If they were really at the same distance as the Coma Cluster (some 320 million light-years)—that is, if the UDGs were in fact part of the cluster—then they were about as large as normal galaxies but also a few hundred times fainter. This implies that they contain at most 1 percent of the expected number of stars for a galaxy of their size.
Follow-up spectroscopic observations of one of the Coma UDGs (Dragonfly 44) with the 10-meter Keck Telescope at Mauna Kea, Hawai‘i, confirmed that the galaxy indeed belongs to the cluster.10 Images obtained with the 8-meter Gemini North Telescope, also at Mauna Kea, further revealed that DF44 is surrounded by many dozens of globular star clusters, just like our own Milky Way, and subsequent velocity measurements yielded a surprisingly large mass, comparable to the mass of our home galaxy. Still, DF44 contains almost no stars. “Dragonfly 44 can be viewed as a failed Milky Way,” the authors concluded in their second follow-up paper, published in 2016.11 With a dark matter fraction of a whopping 98 percent, DF44 apparently represents a new class of “dark galaxies” that has never been recognized before.
The discovery of ultra-diffuse galaxies like DF44 has kept scientists busy over the past few years. No one has a good explanation for the origin of these low-luminosity monsters—they don’t pop up in computer simulations of a ΛCDM universe. Neither does modified Newtonian dynamics offer a satisfactory answer: there are just not enough stars to explain the velocity measurements, even if you use MOND’s alternative gravity equations. As MOND champion Stacy McGaugh says, “DF44 is a problem for everybody.”12 Little wonder that some astronomers question the distance determination, velocity measurements, mass estimate, and even the number of globulars found by the Dragonfly scientists.
Finding “failed” galaxies that are completely dominated by dark matter caused quite a stir, but Dragonfly’s next major discovery was even more controversial. As the team reported in a 2018 Nature publication, they had also hit upon a galaxy that appears to contain hardly any dark matter at all—a faint smudge of light close to the massive elliptical galaxy NGC 1052, which is some 65 million light-years away, much closer than the Coma Cluster.13 From the observed amount of light, van Dokkum and his colleagues derived a stellar (baryonic) mass of some 200 million solar masses, quite typical for a relatively large dwarf galaxy. And like the Coma UDGs, this faint dwarf is surrounded by lots of bright globular star clusters.
Spectroscopic measurements with the Keck Telescope revealed the orbital velocities of ten of those globulars and enabled the astronomers to “weigh” the galaxy, just as they had done with other ultra-diffuse galaxies. However, instead of finding evidence for huge amounts of invisible dark matter, they found a total mass that is hardly larger than the baryonic mass mentioned above. In other words, NGC 1052-DF2, as the enigmatic galaxy is now known, appears to be almost devoid of dark matter.
In 2019 the Dragonfly team announced the discovery of NGC 1052-DF4, a second galaxy in the same group with very similar properties. “The origin of these large, faint galaxies with an excess of luminous globular clusters and an apparent lack of dark matter is, at present, not understood,” they wrote in their Astrophysical Journal Letters paper.14
So first we had an unexplainable deficit of small dwarf galaxies, and dark matter density profiles that are much too flat. Next Dragonfly has given us weird dark matter–dominated galaxies that look like our Milky Way except they contain just 1 percent of the expected number of stars. And now we’re presented with even weirder galaxies that are equally diffuse but appear to be devoid of dark matter altogether. All in all, that’s a lot of inconvenient mysteries, none of which can easily be accounted for by the ΛCDM concordance model of cosmology.
And that’s not all. There’s yet another way in which diminutive galaxies conflict with theoretical predictions and expectations. This final puzzle has nothing to do with the physical or dynamical properties of dwarf galaxies but rather with their three-dimensional distribution in space. Simply put, they’re not where they’re supposed to be.
Detailed supercomputer simulations of the growth of cosmic structure, like IllustrisTNG and EAGLE, show how large galaxies like our Milky Way end up being surrounded on all sides by huge numbers of dark matter subhalos, which become visible as dwarf galaxies. In the real universe, however, the dwarf companions are not only too few in number; they also do not surround their host galaxy equally in every direction. Instead, the majority of satellite galaxies are found in a flattened disk, which does not coincide with the central plane of the host. No matter how computational astrophysicists tweak their code, they are not able to reproduce this distribution in their simulations. It’s known as the planes-of-satellite-galaxies problem.
Back in 1976, when astronomers knew of only eight companions to our Milky Way galaxy (including the Magellanic Clouds), British astrophysicist Donald Lynden-Bell already noticed that most of them approximately line up in one single plane, more or less at right angles with the central plane of the Milky Way. But it wasn’t until 2005 that three European astronomers—Pavel Kroupa, Christian Theis, and Christian Boily—studied the problem in much more detail, by comparing the observed distribution with cold dark matter simulations.15 Their conclusion: the chance of ending up with such a disk-like dwarf galaxy distribution is only 0.5 percent.
Soon enough, it turned out that the planar distribution of Milky Way satellite galaxies is not unique. In 2013 a group led by Rodrigo Ibata of the Observatoire de Strasbourg, France, announced the discovery of a very similar structure around the Andromeda galaxy: about half of Andromeda’s dwarf companions are located in a thin plane some 1.3 million light-years in diameter but only 45,000 light-years thick.16 Moreover, as Ibata’s team showed in their Nature paper, these satellites orbit their host galaxy in the same direction, indicating some common origin or dynamical evolution. And five years later, Swiss astronomer Oliver Müller and his colleagues found that many of the dwarf satellites of the elliptical galaxy Centaurus A, at a distance of 12.5 million light-years, also orbit their massive host in a thin, corotating plane.17
According to Müller’s coauthor Marcel Pawlowski of the Leibniz-Institut für Astrophysik Potsdam, Germany, the planes-of-satellite-galaxies problem has no easy solution.18 “It’s also easy to ignore,” he says, “but by now, many people are at least aware of it.” In 2018, when he was at the University of California Irvine, Pawlowski wrote a comprehensive review article about the problem for Modern Physics Letters A, in which he discussed a number of possible solutions and argued for “why they all are currently unable to satisfactorily resolve the issue.”19
One thing is for sure: if astronomers want to understand the nature of dark matter and its role in cosmic evolution, they have to dig deeper than ever before. Fritz Zwicky focused on the dynamics of massive clusters of galaxies—the largest gravitationally bound structures in the universe. Vera Rubin and Albert Bosma pioneered the study of the rotation of luminous galaxies like our own Milky Way. At these large scales, the presence and influence of mysterious, invisible stuff is very evident. But any viable dark matter theory also needs to fully explain the observed properties and behavior of the unimposing denizens of deep space: the dim satellite dwarfs that swarm around their majestic host galaxies, and the ultra-diffuse “failed” galaxies hiding in the cosmic darkness.
Under the pitch-black New Mexico skies, the sensitive compound eyes of the Dragonfly Telephoto Array may well yield new surprises in the years to come. Van Dokkum and Abraham are expanding their array with more stations, hoping to arrive at a whopping total of 168 lenses. “There’s no reason to stop at forty-eight,” explains van Dokkum. “In principle, you could even build the equivalent of a 10- or 20-meter telescope, at a much lower cost.”
Will cosmologists ever find a way to reconcile their cherished ideas about dark energy and cold dark matter with the observed properties of dwarf galaxies? No one knows, but future observations may provide a solution. Disturbingly enough though, the delinquent dwarfs are not alone in casting shadows on the popular ΛCDM model. Cosmology is facing an even bigger crisis.
22
Cosmological Tension
In the 1980s, when people in East and West Berlin still lived in two very different universes, politically speaking, Checkpoint Charlie was an intimidating and heavily guarded crossing point between communist oppression and liberal democracy. Today it is one of the most popular tourist attractions in the capital of united Germany. But less than three decades after the Berlin Wall was opened in 1989, another insuperable barrier, this time scientific in nature, manifested itself just 600 meters commie-ward of Checkpoint Charlie, in the Auditorium Friedrichstrasse. On a drizzly Saturday in November 2018, this unadorned Soviet-style building served as the intellectual battleground for a cosmological Cold War.
Some 130 scientists had flocked to a one-day symposium to discuss an unnerving crisis in our understanding of the universe.1 During coffee breaks, I ran into a diverse bunch of people from all over the world: astrophysicists and cosmologists, observers and theorists, young postdocs and Nobel laureates. Some of them had spent more time on the plane than they would in the lecture room. Their mutual worry: the universe appears to be expanding too fast, and no one knows why. At the end of the meeting, Brian Schmidt, corecipient of the 2011 Nobel Prize in Physics, told me, “I’m even more puzzled after today.”
Here’s what astronomers and physicists alike were scratching their heads about: working from detailed studies of the cosmic microwave background, the popular ΛCDM model yields a very precise value for the current expansion rate of the universe, with an error margin of just 1 percent. However, “local” measurements, based on observations of galaxies in the relatively nearby universe, arrive at a number that is almost as precise but more than 9 percent higher. And according to Schmidt’s colleague Matthew Colless, one of the organizers of the Berlin symposium, neither side has obvious weak points, even though they can’t both be right.
While some scientists still think there may be an undiscovered error in either one of the two approaches (or maybe in both!), most believe the results are solid. But that doesn’t mean they know how to explain the discrepancy. Even very creative minds like Harvard theorist Avi Loeb are stumped. “I tried to come up with a solution to present at the symposium,” he told the audience, “but I have nothing new to report. It’s not a simple problem to solve.” According to Schmidt, there might be something fundamentally wrong with our interpretation of the cosmic microwave background. Or, who knows, with our current ideas about dark matter.
The determination of the expansion rate of the universe has a history of crisis and controversy. For starters, the earliest guesstimates, back in the 1930s, seemed to indicate that the universe was much younger than the Earth. And just thirty years ago, you’d be offered values that differed by a factor of two, depending on whom you asked. But cosmology has become a high-precision science, and never before has the gap between two different estimates of the Hubble constant—a measure of the current expansion rate—been so statistically significant.
In chapter 15 I explained how the expansion of empty space pushes galaxies away from each other. As a result, all cosmic distances grow by 0.01 percent in 1.4 million years, corresponding to a Hubble constant (or Hubble parameter, usually denoted H 0) of some 70 kilometers per second per megaparsec. But for many decades, the true value of the Hubble constant remained elusive. To determine it, you need to know both the cosmological recession velocity of a galaxy and its distance. In principle the recession velocity, the rate at which a galaxy’s distance is increasing due to the expansion of the universe, can be found by measuring the redshift. But for a nearby galaxy—one for which it’s relatively easy to measure the distance—the redshift measurement is compromised by the galaxy’s real motion through space. These spatial velocities can be as high as a few hundred kilometers per second. As for remote galaxies—the ones for which any spatial motion is negligibly small compared to the cosmological recession velocity—it’s frustratingly hard to measure their distances.
Over the decades, astronomers have found a solution by setting up an elaborate distance ladder to establish how far away other galaxies are. A key ingredient of this technique is a kind of star known as a Cepheid. A Cepheid is a type of variable star: its temperature rises and falls, its diameter expands and contracts, and it brightens and fades. These pulsations occur periodically; the more luminous a Cepheid is, the slower it pulsates. Henrietta Swan Leavitt at Harvard College Observatory discovered this period-luminosity relationship in the early 1900s, and it’s now known as the Leavitt Law. So if you find a Cepheid in another galaxy, its observed period tells you how luminous it is, and the star’s apparent brightness then reveals the galaxy’s distance.
