The BICEP2 Result Was Just Dust, and That’s Okay

Planck on Earth
Figure 1. The Planck satellite being prepared for tests. (Image due to the Planck Collaboration and the European Space Agency.)

You may remember that about this time last year, there was a big hullabaloo because a research group from Harvard claimed that they had discovered primordial gravitational waves using BICEP2, their telescope in the South Pole. This was very exciting because, if true, the result would be extremely good evidence for a model of the early universe called cosmic inflation. (Cosmic inflation is mostly accepted by the scientific community, but it has some philosophical problems and is thus still a little bit controversial. The BICEP2 results would have ended the controversy once and for all.) Even better, the precise strength of the primordial gravitational waves measured by BICEP2 was much greater than anyone had expected, implying that there might be some exciting new things we could learn by studying them further.

Unfortunately, the BICEP2 team’s results were quickly called into question. A competing, space-based experiment, Planck, released some of their own data indicating that what the BICEP2 team had observed was simply cosmic dust. But the analysis wasn’t definitive. The jury was still out.

The beautiful thing about science, though, is that scientists collaborate. The BICEP2 team and the Planck team got together, shared data, and worked through a joint analysis of their measurements. This analysis took several months, but it’s finally been released.

Now we know definitively. The BICEP2 measurement was indeed cosmic dust, not primordial gravitational waves. But the jury is still out on the existence of primordial gravitational waves. It’s just that, if they exist—I personally think they probably do—then they’re as weak as (or weaker than) we originally thought, not as strong as the BICEP2 measurement indicated.

But why did BICEP2 get such a strong false positive? How did their measurement go so wrong? Well, hold on to your hats, ladies and gentlemen, because I’m going to explain.

This is a follow up to my post from last March on the BICEP2 result. If you haven’t read that yet, I suggest you go check it out before continuing.

A Much-Too-Short Summary of Cosmic Inflation and the CMB

About 13.8 billion years ago, the universe was extremely hot, so hot that matter couldn’t form at all… it was just a chaotic soup of charged particles. Hot things (and accelerating charges) glow. And this hot soup was glowing incredibly brightly. As time passed, the universe expanded and cooled, but this glow remained, bathing all of time and space in light.

(The reason for why the universe was so hot in the first place depends on whether cosmic inflation is true. Either it’s because the Big Bang just happened or it’s because, after cosmic inflation, a particle called the inflaton dumped all of its energy into creating hot matter.)

Even today, the glow remains, filling the universe. As the universe expanded, the glow dimmed and its light changed colors (due to gravitational redshift), until it became microwaves instead of visible or ultraviolet light. This ubiquitous glow is called the Cosmic Microwave Background, or CMB for short, and if you turn an old analogue TV to an unused channel, some of the static you hear is CMB radiation picked up by your TV antenna.

Since its discovery, the CMB has been one of our most powerful probes of cosmology. It lets us accurately measure how fast the universe is expanding, the relative amounts of normal stuff vs dark energy and dark matter, how the density of matter fluctuated in the early universe, how the Earth is moving relative to the expansion of the universe, and much more.

Measuring the CMB

One amazing thing about the CMB is that all of the light that reaches us is the same color, to an incredible degree. However, the color does fluctuate a little bit…in a special way that’s independent of position in the sky or direction. These tiny deviations from the norm are primarily what we’d like to measure.

So how does a measurement work? How can we measure something that’s literally everywhere? On Earth, we can see things in three dimensions because we have two eyes separated from each other. But on the scale of the CMB, which fills the entire universe, the whole Earth looks like a single eye–in other words, from our perspective, the sky is two-dimensional.

CMB-incoming
Figure 2. Although light from the CMB (red) fills the universe, when we observe it on Earth, it looks like it’s been projected onto a two-dimensional screen (blue) above our heads. (Earth courtesy of NASA.)

This means we observe all of the light from the CMB as if it were projected onto a spherical screen above our heads, as shown in Figure 2. Looking from the outside in, the result is something like Figure 3, which plots the wavelength of the light across the sky. (The differences in the wavelengths have been enhanced by about a factor of a million.)

the CMB on our sky
Figure 3. The cosmic microwave background projected onto our sky, looking from the outside in. (Image made by Damien P. George with data from Planck.)

Of course, although three-dimensional models are easiest to visualize, they’re not great to actually work with. So we usually map the CMB onto a flat surface, the same way we map the Earth. This is what gives rise to the famous “all-sky” maps like the one shown in Figure 4.

Planck_CMB
Figure 4. The measured CMB mapped on a flat surface. (Image due to the Planck collaboration.)

There’s a lot of information hidden in Figure 4 that you can’t see unless you do some serious math. In fact, you could learn almost everything I’ve told you so far just from looking at the CMB! And there’s more to learn as we make new and increasingly precise measurements.

Planck Vs. BICEP

It’s at this point that I need to provide a clarifying comparison. The images I just showed you were generated by the Planck satellite, which is a small satellite that lives just beyond the moon’s orbit. As Planck orbits the Earth (and as the Earth orbits the sun), it makes measurements of the CMB in small segments of the sky. Over the course of a year, it can build up a map of the CMB in the entire sky, as shown in Figures 3 and 4. (Planck also takes measurements of several different wavelengths of light and aggregates the data. This is important and I’ll get back to that.)

BICEP2 (shown in Figure 5),  on the other hand, is a single telescope near the South Pole. The BICEP2 people chose to measure a small patch of the sky extremely precisely and they only measured one wavelength of CMB light.

BICEP2 sunset
Figure 5. The BICEP2 telescope at the South Pole. (Courtesy of the BICEP collaboration).

What Went Wrong?

If sending a satellite into space or pointing a telescope at the sky were all that was required to precisely measure the CMB, the BICEP2 team never would have mistaken dust for gravitational waves. So what went wrong?

Well, I told you that the fluctuations in the CMB are very very small. This means that they can be drowned out by the many other sources of microwaves in the universe. Jupiter, the sun, black holes, pulsars, cosmic dust…tons of things produce microwaves. Collectively, all this other stuff is called foreground.

To screen out the foreground, cosmologists build an extremely detailed map of non-CMB sources of microwave radiation in they sky, called a mask, and subtract it from the map of microwave light that the instrument actually measured. After the subtraction, you get something like Figure 4. The mask used to remove known sources in the Milky Way is shown in Figure 6.

Planck Galaxy Mask (for the Power Spectrum)
The mask that the Planck collaboration used to screen out galactic microwave sources from their measurements of the CMB. (Source: the Planck collaboration)

But mask-making is tricky business. To build a map, cosmologists use previous measurements of the sky and computer simulations. The Planck collaboration uses an additional trick: they can detect several different wavelengths of microwaves. The only microwave source that will look the same in every wavelength is the CMB, so by comparing the measurements in different wavelengths, Planck can remove unexpected sources of noise.

But BICEP2 only measured one wavelength of light, and this is what killed it. The computer models the BICEP2 people used to make a mask for their little corner of sky didn’t predict that it contained as much spinning cosmic dust as it does. Planck, with their multi-wavelength detector, wasn’t fooled in the same way.

(I should emphasize that the BICEP2 team’s mask was flawed. The team based their dust estimates on older measurements and made a mistake when estimating how much the radiation from the dust would change when you looked at a different color of light. But these are subtle errors, and having several colors of light to look at would have been a fail-safe against them.)

BICEP2 Didn’t Do Anything Wrong

It’s tempting to say that the BICEP2 collaboration failed in some way—their data analysis was poor, they designed their experiment badly, etc. But they couldn’t have known that this cosmic dust would have been a problem. It’s easy to see what to do in hindsight…not so much when you’re planning a multimillion-dollar project years (or even decades) in advance.

This is how science is done. We make a prediction, we design an experiment, we measure something in the world, and we invariably mess up. But by keeping our minds open to our own fallibility, we give ourselves the opportunity to try again and eventually get it right. That’s what happened here. BICEP2 made an erroneous conclusion, took the opportunity to collaborate with Planck, and they figured it out.

That’s fantastic. That’s what I call science.

(Parenthetical note. There are other ways that the BICEP2 team deserves criticism. Before submitting their article to peer review, the team held a huge press conference and generated a lot of publicity. Given that the conclusion wasn’t yet vetted by the scientific community, this kind of behaviour can and probably did detract from the credibility of science in the public’s eye.)

Further Reading

  • If you’d like to read the joint Planck-BICEP2 press release, you can find it here. In the press release, there’s an link to the scientific paper that the two collaborations wrote together, which is currently undergoing peer review.
  • Planck periodically releases their measurements and statistical tools, including software, masks, and all-sky maps, to the scientific community at large. This is where I got the galaxy mask I showed. If you’d like to browse, you can find all the data and documentation explaining how to use it here.
  • If you’re very brave, you can read this review paper on how foregrounds are removed from CMB measurements. This article describes in great detail how masks are generated.
  • If you want to look at the CMB in its full spherical glory, Damien P. George created this webapp. It’s pretty awesome.

Related Reading

  • This article is a follow-up to my previous article on BICEP2, which you can find here.
  • If you’d like to know how we know that the universe is expanding, you might want to check out my article on exactly that.

If you’re confused about the Big Bang or this whole “inflation” thing, you might want to check out my three-part series on the early universe:

Acknowledgements

Thanks to Alexandra Fresch for proof-reading and editing and thanks to Sara Simon for making sure I get the cosmology right.

3 thoughts on “The BICEP2 Result Was Just Dust, and That’s Okay

Comments are closed.