Faultory

Case file 83 · Science & Discovery

A twist in the oldest light

Cosmic inflation, 2014: did a polarization signal reveal gravitational waves, or dust in our own galaxy?

Subject
BICEP2 gravitational wave claim (2014)
Reading
15 min
Evidence
5 exhibits
Reasoning
3 steps
Level
Advanced
A laboratory building and telescope enclosures at the South Pole station.
Photographed 18 January 2011 The Dark Sector Laboratory at the South Pole station, with the BICEP2 telescope at right, photographed before the 2014 announcement. Photo: Ketiltrout · CC BY-SA 3.0 · Wikimedia Commons.
Illustrative signature. An illustrative strip of polarization readings shows a small oscillating swing riding on noise (a sketch, not data). Drawn to explain the case, not plotted from source data.

01 · The briefing

Begin with the question.

The cosmic microwave background is the afterglow of the early universe, and its light carries a faint polarization, a preferred orientation in the way its waves vibrate. Polarization patterns come in two types. The B-mode pattern is the one that theory links to primordial gravitational waves produced by cosmic inflation, although gravitational lensing by intervening matter can also produce B-modes.

In March 2014 the BICEP2 team, observing from the Amundsen-Scott South Pole Station, reported a B-mode excess at degree angular scales. The team's model gave a tensor-to-scalar ratio r of 0.20, with the no-tensor case disfavored at a stated 7.0 sigma. The result rested on one experiment, so other data sets were the obvious check. The BBC reported that the discovery had been hailed as a near-certain route to a Nobel Prize. Within months, critics pointed out that polarized emission from dust in the Milky Way can mimic the same pattern.

Three explanations were therefore in play: primordial gravitational waves, polarized galactic dust, or an instrumental artifact. This investigation asks which of them the later evidence supports, and how much of the 2014 signal the data can actually attribute to inflation.

Several telescopes and observatory buildings at the South Pole.
Where it happenedPhotographed 7 January 2011 The Dark Sector at the South Pole, with the BICEP2 telescope to the right of the laboratory that houses the South Pole Telescope. Photo: Ketiltrout · CC BY-SA 3.0 · Wikimedia Commons.

The timeline

  1. The BICEP2 team announces a detection of B-mode polarization at degree scales, with r = 0.20 and r = 0 disfavored at 7.0 sigma in its model.

  2. The BICEP2 paper is formally published in Physical Review Letters.

  3. The BBC reports that the team's confidence has been lowered as critics focus on whether dust could account for the signal.

  4. The BICEP2 paper appears in a revised arXiv version that still reports r = 0.20 and the 7.0 sigma figure, while critics continue to test dust as the explanation.

  5. The Planck collaboration posts an analysis of polarized dust emission that, extrapolated to the BICEP2 frequency, gives a dust power of the same magnitude as the BICEP2 excess.

  6. The BICEP2/Keck and Planck collaborations post a joint analysis that finds strong evidence for dust and no statistically significant evidence for tensor modes.

A researched reconstruction for learning. These reasoning steps simplify a complex record; no score can capture the full history.

02 · The evidence

What does the record show?

A scale model of a spacecraft.
On the evidence tablePhotographed 14 February 2012 A scale model of the Planck spacecraft, whose data were used to estimate dust in the BICEP2 field. Photo: Mike Peel · CC BY-SA 4.0 · Wikimedia Commons.

Read each exhibit before settling on an explanation. The source links below every exhibit lead to the record behind it.

Exhibit 01

The BICEP2 fit and its significance

The BICEP2 paper reports a fitted tensor-to-scalar ratio of r = 0.20, with r = 0 disfavored at 7.0 sigma under its lensed-LCDM plus tensor model. In the same abstract the authors state that accounting for foreground dust will shift this value downward. The paper establishes the observed B-mode excess and its significance within that model. It does not establish that the excess is primordial, because the fitted value depends on how dust is treated.

Source record s1arxiv.org

Exhibit 02

The team's own limit on dust models

The BICEP2 paper examines available models of polarized dust and finds that, at their default parameter values, they predict power about 5 to 10 times smaller than the observed excess. The authors add that these models are not sufficiently constrained by external public data to exclude dust emission bright enough to explain the whole excess. The same abstract reports that a cross-correlation with BICEP1 data at 100 GHz disfavours dust at 1.7 sigma, a weaker test. The passage establishes that dust was not excluded by the team's own comparison. It does not establish that dust is bright enough, which is the question the later analyses address.

Source record s1arxiv.org

Exhibit 03

Planck's dust level in the same field

The Planck paper extrapolates its 353 GHz polarized dust data to 150 GHz and gives a dust power of 1.32 times 10 to the minus two microkelvin squared over the multipole range 40 to 120, with a statistical uncertainty of 0.29 and an extrapolation uncertainty of +0.28 and -0.24 in the same units. The authors state that this level is the same magnitude as the BICEP2 signal over that range. The paper establishes a dust level in the targeted field by extrapolation. It does not measure dust directly at the BICEP2 frequency.

Source record s2arxiv.org

Exhibit 04

The joint analysis of BICEP2, Keck and Planck

The 2015 joint analysis reports that the data give strong evidence for dust and no statistically significant evidence for tensor modes. It gives an upper limit of r_0.05 below 0.12 at 95 percent confidence, and it reports lensing B-modes detected at 7.0 sigma after marginalizing over dust and r. The paper establishes the revised constraint. The r limit depends on the dust model and its frequency prior, and it is not a measurement of zero primordial signal.

Source record s3arxiv.org

Exhibit 05

A news account of the criticism

A BBC News report from June 2014, titled Cosmic inflation: Confidence lowered for Big Bang signal, says the critiques since March had focused on whether polarized dust could mimic the pattern, and quotes a team member saying his confidence had gone down. It records the public mood and the timeline of criticism. It is a secondary account of the scientific debate and does not establish the physics.

Source record s4bbc.com

03 · Your reasoning

Follow the strongest explanation.

Choose an interpretation, then compare it with the evidence. You can revise your choice. The goal is understanding, not a perfect score.

Reasoning step 01 of 03

Which explanation best fits the evidence after the 2015 joint analysis?

Evidence-based explanation

The joint analysis of BICEP2/Keck and Planck data finds strong evidence for dust and no statistically significant evidence for tensor modes. Its upper limit of r below 0.12 at 95 percent confidence is much weaker than the 2014 claim. The primordial reading is not excluded, but the data do not demonstrate it, and the result depends on the dust model and its frequency prior.

Weigh it against

Exhibit 01The BICEP2 fit and its significanceExhibit 03Planck's dust level in the same fieldExhibit 04The joint analysis of BICEP2, Keck and Planck

Reasoning step 02 of 03

What does the 7.0 sigma figure in the 2015 joint analysis refer to?

Evidence-based explanation

The 7.0 sigma figure in the 2015 analysis is reported for lensing B-modes after marginalizing over dust and r. It is not a detection of primordial tensor modes, which the same analysis does not find significant. Readers often carry the 2014 significance figure into the 2015 result, so the wording of each claim matters.

Weigh it against

Exhibit 01The BICEP2 fit and its significanceExhibit 04The joint analysis of BICEP2, Keck and Planck

Reasoning step 03 of 03

Which measurement would most directly separate primordial polarization from dust?

Evidence-based explanation

Dust and the primordial signal have different frequency dependence, so a measurement at a second frequency gives a direct test that does not rely on modeling one component alone. The Planck work and the joint analysis both rely on the frequency behavior of dust for this reason. The test is only as good as the dust model at the other frequency, which is why the joint result was reported as a limit.

Weigh it against

Exhibit 02The team's own limit on dust modelsExhibit 03Planck's dust level in the same fieldExhibit 04The joint analysis of BICEP2, Keck and Planck

04 · The findings

The chain behind the outcome.

The BICEP2 team reported in March 2014 a B-mode polarization excess at degree scales, fitted with a tensor-to-scalar ratio of 0.20 and a 7.0 sigma rejection of r = 0 within its model. That is a measurement of the observed excess. Its interpretation depended on how much of the excess was foreground dust, and the team's own paper said that the dust models it examined predicted power about five to ten times smaller than the excess while not being constrained well enough to exclude dust that could explain all of it.

Planck's analysis of polarized dust, posted to arXiv in September 2014, extrapolated to the BICEP2 frequency and gave a dust power of the same magnitude as the BICEP2 signal, with a sizeable extrapolation uncertainty. The 2015 joint analysis of BICEP2/Keck and Planck data then found strong evidence for dust and no statistically significant evidence for tensor modes, with an upper limit on r below 0.12 at 95 percent confidence.

The best-supported reading is therefore that the 2014 excess was substantially contaminated by, and may have been dominated by, polarized galactic dust, and that the data did not establish primordial gravitational waves. The revision happened in public, through independent analysis of the same data and a joint analysis that adopted the critics' central point.

Contributing causes and conditions

  1. Primordial gravitational waves, if present, would leave a B-mode pattern at degree scales in the polarization of the cosmic microwave background.
  2. Foreground dust in the Milky Way emits polarized light that can mimic a B-mode pattern at similar angular scales.
  3. Dust brightness depends on frequency in a way that differs from the primordial signal, so measurements at several frequencies can separate them.
  4. The 2014 fit assumed a dust model whose brightness was not constrained well enough by external data to rule out a dust-dominated excess.
  5. The 2015 joint analysis combined BICEP2/Keck and Planck data with a prior on dust's frequency behaviour from Planck's analysis of other sky regions, and found strong evidence for dust and no statistically significant evidence for tensor modes.

OutcomeBICEP2 gravitational wave claim (2014)

Take the method with you

  1. A headline significance figure depends on the model behind it, so ask which components were marginalized or fixed.
  2. A critic's alternative should make a different prediction, ideally at another frequency, rather than only a different story.
  3. A revised limit from a joint analysis is real progress, even when it reverses an earlier headline.

You can read findings at any time. Exploring every reasoning step marks this case as complete in your notebook.

Open another investigation

The open record

Sources & further reading

Read the original record and its scope before relying on a conclusion. Source references support the exhibit summaries; Faultory’s reasoning prompts are editorial reconstructions.

  1. s1
  2. s2
  3. s3
  4. s4

Written and curated for Faultory by Yunus Emre Vurgun, with AI-assisted research and drafting. Sources and interpretations can need correction. Read our editorial standards or report an issue.

Keep investigating