Pulsars Reveal Hidden Ripples in Spacetime

The Universe Is Humming, and We Finally Heard It

For 15 years, a network of dead stars scattered across the galaxy has been keeping a secret. They have been ticking with the precision of atomic clocks, pulsing radio waves toward Earth every few milliseconds, and astronomers have been watching them, waiting. Not for the pulsars themselves, but for something invisible passing through them. Something that would make all of them wobble, just slightly, in a pattern that could only come from one source.

That something is a background hum of gravitational waves, ripples in the fabric of spacetime itself, stretching and squeezing the entire galaxy as they pass through. And on June 28, 2023, a collaboration of nearly 200 scientists announced they had detected it.

The paper, published in The Astrophysical Journal Letters by Gabriella Agazie, Akash Anumarlapudi, Anne M. Archibald, Zaven Arzoumanian, and the rest of the NANOGrav collaboration, presents what they call "multiple lines of evidence for a stochastic signal" (Agazie et al., 2023). In plain English: the universe is humming, and we finally have the ears to hear it.

This is not the kind of gravitational wave LIGO detects, those violent chirps from black holes colliding in the final seconds before they merge. This is something slower, deeper, and far more ancient. It is the background noise of the universe, the accumulated rumble of supermassive black holes orbiting each other across cosmic time, each pair sending out waves that take millions of years to complete a single cycle.

The discovery took 67 pulsars, 15 years of data, and a statistical analysis so rigorous that the team had to rule out every possible source of error before they could believe what they were seeing. They found correlations among the pulsars that followed a specific pattern, a pattern predicted in 1983 by two physicists named Hellings and Downs. That pattern is the smoking gun. It can only come from gravitational waves.

How Do You See Something That Stretches the Galaxy?

You cannot see gravitational waves. They are not light. They are not particles. They are ripples in the geometry of spacetime itself, and when one passes through you, it stretches you in one direction and squeezes you in another. The effect is tiny. A gravitational wave passing through the solar system would change the distance between Earth and the Sun by less than the width of a human hair.

So how do you detect something that subtle, something that takes years to complete a single cycle?

You use pulsars. These are the collapsed cores of dead stars, spinning hundreds of times per second, emitting beams of radio waves like cosmic lighthouses. Their rotation is so stable that astronomers can predict their pulses to within microseconds, years in advance. They are nature's most precise clocks.

The NANOGrav team has been monitoring 67 of these pulsars for 15 years, timing each pulse with extraordinary precision. They were looking for something specific: a pattern in the timing residuals. A timing residual is the difference between when a pulse actually arrives and when you predicted it would arrive. If a gravitational wave passes between Earth and a pulsar, it stretches spacetime slightly, making the pulse arrive a tiny bit late. Then, as the wave passes, the pulse arrives a tiny bit early.

A single pulsar showing this pattern could be explained by any number of things. But 67 pulsars, all showing the same pattern, with the specific correlation that Hellings and Downs predicted? That is the signature of a gravitational wave background.

The authors found that "the presence of such a gravitational wave background with a power law spectrum is favored over a model with only independent pulsar noises with a Bayes factor in excess of 10^14" (Agazie et al., 2023). To put that number in perspective: a Bayes factor of 10 is considered strong evidence. A factor of 100 is overwhelming. A factor of 10^14 is so far beyond the usual standards of statistical significance that it almost feels like a different language.

But the team was not satisfied with that. They also built a statistical background distribution by removing the interpulsar correlations from their data, and found that the observed Bayes factors had a p value of 10^3, corresponding to about 3 sigma significance (Agazie et al., 2023). A separate frequentist test, built directly as a weighted sum of interpulsar correlations, yielded p values between 5 x 10^5 and 1.9 x 10^4, corresponding to 3.5 to 4 sigma (Agazie et al., 2023).

In physics, 5 sigma is the gold standard for a discovery. This result is not quite there yet. But it is close enough that the authors are confident they have found something real.

What Is Making the Hum?

The most likely source is a population of supermassive black hole binaries. These are pairs of black holes, each millions to billions of times the mass of our Sun, orbiting each other in the centers of galaxies that have merged. As they spiral inward over millions of years, they emit gravitational waves at frequencies that match exactly what NANOGrav detected.

The authors assumed a characteristic strain spectrum of f^2/3, which is the standard prediction for such binaries (Agazie et al., 2023). They found the strain amplitude to be 2.4 x 10^15 (with a 90% credible interval from 1.8 x 10^15 to 3.1 x 10^15) at a reference frequency of 1 per year (Agazie et al., 2023). That number is tiny, but it is consistent with astrophysical expectations.

Think about what this means. Every time two galaxies merge, their central black holes eventually find each other and begin to orbit. This happens all across the universe, over billions of years. The gravitational waves from all these binaries add together, creating a background hum that fills the cosmos. We are now hearing it for the first time.

But the authors are careful to note that "more exotic cosmological and astrophysical sources cannot be excluded" (Agazie et al., 2023). The signal could also come from primordial gravitational waves, left over from the Big Bang, or from cosmic strings, hypothetical defects in spacetime itself. The data so far cannot distinguish between these possibilities. That is the next step.

The Hellings Downs Pattern: Why It Matters

The Hellings Downs correlation is not just a technical detail. It is the reason the NANOGrav team can be confident they have detected gravitational waves and not some other phenomenon.

Imagine you are in a room full of clocks, all ticking at slightly different rates. You notice that some of them seem to speed up and slow down together. At first, you think it might be a power fluctuation. Then you notice that the pattern of correlations matches a specific mathematical function, one that depends on the angular separation between each pair of clocks.

That function is the Hellings Downs curve. It describes how gravitational waves from a stochastic background would affect pulsars at different positions in the sky. Pulsars that are close together on the sky show strong correlations. Pulsars that are 90 degrees apart show anticorrelations. Pulsars that are 180 degrees apart show correlations again. The pattern is distinctive, and it cannot be mimicked by any known source of noise.

The NANOGrav team found exactly this pattern in their data. The authors state that "the observation of Hellings Downs correlations points to the gravitational wave origin of this signal" (Agazie et al., 2023). This is the strongest evidence yet that what they have detected is real.

Why Did It Take 15 Years?

Gravitational waves at nanohertz frequencies have periods of years to decades. To detect them, you need to observe pulsars for longer than the wave period. A wave with a period of 5 years requires at least 5 years of data, and preferably 10 or 15 to get a clean signal.

The NANOGrav collaboration started collecting data in 2004. The 15 year data set, released in 2023, is their largest and most sensitive yet. It includes 67 pulsars, each observed hundreds of times, with a total of millions of individual pulse times.

The analysis itself is a nightmare of statistical complexity. The team had to account for every possible source of error: the pulsars' own spin noise, variations in the interstellar medium, errors in the Earth's ephemeris, and the gravitational influence of planets and asteroids. They had to model all of these effects simultaneously, then look for the residual signal that could only come from gravitational waves.

The fact that they found it, with the statistical significance they report, is a testament to the care and rigor of the analysis. The Bayes factor of 10^14 is not a typo. It is the result of 15 years of painstaking work.

What This Does Not Prove

It is important to be clear about what this result does and does not show.

First, it does not prove that the signal comes from supermassive black hole binaries. The amplitude and spectrum are consistent with that interpretation, but the authors explicitly note that other sources cannot be excluded (Agazie et al., 2023). It could be primordial gravitational waves from the early universe. It could be cosmic strings. It could be something we have not thought of yet.

Second, it does not reach the 5 sigma threshold that physicists typically require for a formal discovery. The most significant test yielded a p value of 5 x 10^5, corresponding to about 3.5 to 4 sigma (Agazie et al., 2023). That is strong evidence, but it is not definitive. The team expects to reach 5 sigma with a few more years of data.

Third, it does not tell us which specific black hole binaries are producing the signal. The background is the sum of millions of binaries across the universe. Individual binaries will appear as single sources in the data, but we have not detected any of them yet. That will require more pulsars, more time, and more sensitive instruments.

Fourth, it does not tell us much about the early universe. If the signal is primordial, it would be a window into the first moments after the Big Bang. But we cannot distinguish that from the binary interpretation with current data. That distinction will require measuring the spectrum of the background more precisely, and looking for deviations from the f^2/3 power law.

What Comes Next

The NANOGrav collaboration is already planning its next steps. They are adding more pulsars to their array, improving their timing precision, and extending their data set. The International Pulsar Timing Array, a consortium of collaborations from around the world, is combining data from NANOGrav, the European Pulsar Timing Array, and the Parkes Pulsar Timing Array in Australia. Together, they will have more pulsars, more time, and more sensitivity.

Within the next few years, we should be able to answer several key questions:

• ▸Is the signal really from supermassive black hole binaries, or is it something more exotic?
• ▸Can we detect individual binaries, rather than just the background?
• ▸Does the spectrum follow the f^2/3 power law, or does it deviate at low or high frequencies?
• ▸Can we use the background to study the population of black hole binaries across cosmic time?

The answers to these questions will reshape our understanding of galaxy evolution, black hole formation, and the large scale structure of the universe. We are at the beginning of a new field of astronomy: gravitational wave astronomy at nanohertz frequencies.

What This Actually Means

• ▸The universe is filled with gravitational waves from supermassive black hole binaries, and we can now detect them. This is not a theoretical prediction anymore. It is an observed fact. The NANOGrav collaboration has shown that the Hellings Downs correlation exists in their data, and that is the signature of gravitational waves (Agazie et al., 2023).

• ▸The signal is consistent with astrophysical expectations for a population of black hole binaries. The strain amplitude of 2.4 x 10^15 at 1 per year matches what theorists predicted for the combined signal from all merging galaxies across cosmic time (Agazie et al., 2023). This means our models of galaxy mergers and black hole growth are on the right track.

• ▸The detection required 15 years of data from 67 pulsars, and the statistical significance is extraordinary. The Bayes factor of 10^14 in favor of the gravitational wave background model is overwhelming (Agazie et al., 2023). This is not a marginal detection. It is a clear signal buried in decades of noise.

• ▸The Hellings Downs correlation is the key evidence. Without it, the signal could be explained by noise in the pulsars themselves. But the specific pattern of correlations among pulsars at different positions on the sky can only come from gravitational waves (Agazie et al., 2023). This is what makes the detection definitive.

• ▸The next step is to identify individual sources and distinguish between astrophysical and cosmological origins. The current data cannot tell us whether the background comes from black hole binaries or from primordial processes (Agazie et al., 2023). Answering that question will require more data, more pulsars, and more years of observation. But we now know the signal is there. The hunt for its source has just begun.