Finding the Source of the Pioneer Anomaly

Thirty years ago, the first spacecraft sent to explore the outer solar system started slowing unexpectedly. Now we finally know what happened

14 min read
Illustration: noemotion
Illustration: noemotion

Some 40 years ago, a quarter-ton lump of circuits and sensors slipped Earth’s surly bonds, sped past the moon and Mars, and hurtled toward Jupiter. The probe, Pioneer 10, and its sister ship, Pioneer 11, which followed a year later, were true trailblazers. They gave humanity its first close-up glimpses of worlds beyond the solar system’s asteroid belt. They also left behind a mystery—one that has simultaneously baffled and inspired astrophysicists for years.

Like many puzzles, this one started out with just a small hint that something was amiss. Not long after Pioneer 10 and 11 had passed beyond the orbits of Jupiter and Saturn, their navigators began to notice something unexpected. Both spacecraft seemed to be slowing down more than controllers had predicted they would, as if some force were tugging them ever so subtly backward toward the sun.

The magnitude of this deceleration was minuscule, just one ten-billionth of the gravitational acceleration that we experience on Earth’s surface. Such a small effect didn’t seem out of place at first. It would have shown up as a simple correction, the sort that spacecraft navigators routinely apply to accommodate fuel leaks and other small, transient deviations in spacecraft behavior. No one would have blinked an eye if it hadn’t been for one troubling detail: For years, as the spacecraft sped deeper and deeper into space, that tiny discrepancy stuck around. And no one could figure out where it came from.

When the discovery of this strange anomaly was finally announced in 1998, more than 15 years after it was first detected, it didn’t take long for potential explanations to emerge. Studies of distant stellar explosions had just shown that the universe is expanding at an accelerating rate. The magnitude of the Pioneers’ deceleration was in just the right ballpark to suggest that we might be seeing a manifestation of this same expansion within our own solar system. But that was just one possibility. Some suspected the two spacecraft were being tugged by an unseen cloud of dark matter or else were exposing a breakdown in the laws of gravitation at large distances. Others thought the spacecraft might have unearthed the first evidence of extra dimensions or exotic “mirror matter.”

All told, hundreds of papers have been written that attempted to explain the Pioneer anomaly with new physics. But none of the proposed explanations has been entirely satisfactory, and there has always been the possibility that some more mundane mechanism—perhaps some aspect of the spacecraft themselves—was responsible. Now, after a dedicated search for lost spacecraft records, detailed study of onboard sensor and navigational data, and careful modeling of the spacecraft and their trajectories, we believe we finally know the answer. Our effort to identify the source of this anomaly is in part a detective story and part a cautionary tale about the importance of preserving seemingly uninteresting data.

When it comes to understanding the Pioneer anomaly, it helps to start more or less at the beginning. When Pioneer 10 and 11 were conceived in the mid-1960s, no spacecraft had yet been sent past the orbit of Mars. No one knew how risky it would be to cross the asteroid belt between Mars and Jupiter or how difficult it would be to withstand the zones of intense radiation that encircle the gas giant. But Pioneer 10 and 11’s designers nevertheless reckoned they had what they needed to make these missions work.

Out by Jupiter, the intensity of sunlight is about 4 percent of what it is in Earth orbit, far too dim to power the solar panels of the day. So each spacecraft carried a set of four radioisotope thermoelectric generators (RTGs), positioned a safe distance away from the spacecraft’s sensitive instruments on two long booms, a bit like a pair of rabbit ears. These generators converted heat from the radioactive decay of plutonium-238 pellets directly into electricity using bimetallic thermocouples. The RTGs were ultrareliable: They contained no moving parts, and the 88-year half-life of the fuel meant the spacecraft would have adequate power for decades.

To communicate with Earth, each spacecraft was equipped with a 2.74-meter-wide parabolic antenna. It was decided early on that the probes would be stabilized by spinning them along the axis of their antennas. This kept each spacecraft’s antenna pointed in a consistent direction, cutting down on the need for frequent attitude corrections. If an antenna were to fall out of alignment with Earth, incoming signals would hit the antenna off-axis. Special circuits would then pick up on the modulation of the signal that would arise as it wobbled back and forth across the antenna while the spacecraft rotated. These circuits would then fire hydrazine thrusters to correct the spacecraft’s orientation.

Armed with this basic technology, Pioneer 10 and 11 set off where no man or probe had gone before, on trajectories that would eventually take them to nearly opposite ends of the solar system. After its launch on 2 March 1972, Pioneer 10 traversed the asteroid belt and swung past Jupiter in December 1973, its off-axis single-pixel photodetector spinning slowly to construct pictures of the giant planet. Pioneer 11, which launched on 6 April 1973, performed an even more ambitious mission: flying past Jupiter a year after Pioneer 10, and then using the gas giant’s gravity to change course—performing one of the first of the now-common “slingshot maneuvers.” The spacecraft flew beneath Saturn’s rings some five years later.

The probes, of course, continued hurtling through space after their planetary flybys. As Earth receded into the distance, our planet’s apparent motion around the sun diminished as well. Consequently, the thrusters needed to fire less and less often every year to adjust the orientation of Pioneer 10 and 11’s homeward-directed antennas. That turned the spacecraft into largely ballistic objects, and as such, nearly ideal probes of the effect of external forces like gravity.

At this point, a group of researchers led by astronomer John D. Anderson at NASA’s Jet Propulsion Laboratory (JPL) set out to use navigational data on these mostly passive probes to make high-precision measurements of the gravitational environment of the outer solar system. The team, which included one of us (Turyshev), was specifically looking for subtle changes in trajectory that might be caused by the tug of as‑yet-undiscovered planets or else very-long-wavelength gravitational waves produced during the big bang. Neither signal turned up. Instead, the investigation uncovered something completely unexpected and far harder to interpret.

To use Pioneer 10 and 11 as gravity probes, Anderson’s team relied primarily on signals that were originally sent from Earth using NASA’s Deep Space Network, a collection of large radio antennas located at three sites around the world. These antennas sent out a 2.11-gigahertz radio signal, which was received by the Pioneer spacecraft. The spacecraft, which did not have their own precision oscillators, used the frequency of the incoming beam to tune their return signals to 2.29 GHz.

The well-known Doppler effect (of passing-train fame) says that the received frequency of a radio signal changes if the source and the receiver are moving relative to each other. By measuring how far a Pioneer signal shifted from 2.29 GHz, ground operators could tell what the spacecraft’s line-of-sight velocity was with respect to Earth. After that, it was all a matter of time and analysis. Repeated measurements of line-of-sight velocity over a sufficiently long arc of a spacecraft’s trajectory could be used to constrain its path.

Of course, to tease out any unknown effects, the team had to think of everything that could alter the probes’ motion or the frequency of the signals they sent. JPL navigators had to build a model that took into account, among other things, the gravitational influence of all known solar-system bodies, solar weather, solar radiation pressure, the influence of general relativity on the time of flight of signals, excess moisture in Earth’s atmosphere above the transmitting or receiving station, and the precession of Earth’s pole.

When these factors were fed into the model of each probe’s trajectory, the team expected to see their received radio signals match within a few millihertz of predictions. But they didn’t. Beginning in 1980, Anderson and his colleagues noticed a much more significant deviation—on the level of a few tens of millihertz. It quickly became clear that the discrepancy was not just random noise.

The accuracy of the navigational model could be restored if a small, constant sunward force was presumed to be acting on the spacecraft. But the team had no ready explanation for such a force in the realm of known physics. When discovery of the anomaly was finally announced in the pages of Physical Review Letters in 1998, theoretical physicists quickly responded with proposals. But there were lingering doubts. It was still quite possible that some mundane physical explanation, such as leaky thrusters, might account for the force.

Thermal radiation was a leading candidate. The Pioneers’ RTGs were very inefficient. The generators supplied roughly 100 watts (more at launch) of electricity to each probe’s systems, but in doing so they radiated a good 2.5 kilowatts of heat into space. The slight deceleration experienced by the Pioneers could be explained away by a tiny asymmetry. If only 60 W more heat had radiated off the back of each probe, behind the antenna, conservation of momentum would cause them to recoil just enough in the other direction, toward Earth, to explain the anomaly.

To follow up, the JPL team spent the next several years preparing a comprehensive study. They processed more than 10 years’ worth of Doppler measurements from Pioneer 10 and nearly four years of similar data from Pioneer 11. They compiled a thorough inventory of all possible sources of error and bias, including onboard heat. Ultimately, they produced a paper in 2002 that ran nearly 60 pages (one of the longest articles ever published in Physical Review D) addressing all these questions in detail.

The team concluded that thermal recoil was probably not responsible for the Pioneer anomaly, for two reasons. One, because the RTGs were extended from the body of each spacecraft on 2.5-meter-long booms, it didn’t seem likely that there would be enough excess heat bouncing off the craft (the team estimated it was likely just one-tenth of what was needed to account for the effect). Two, the deceleration of both spacecraft seemed constant in magnitude, even decades into the mission. If the RTGs were responsible for the Pioneer anomaly, the team argued, the effect should have diminished in magnitude as the Pu-238 decayed and produced less power, and thus, less heat.

But nagging doubts remained. The issue of the thermal-recoil force was answered using relatively crude, back-of-the-envelope estimates. And while a constant push seemed to fit the anomaly very well, no one could really exclude the possibility that it was in reality diminishing over time.

How could these questions be answered? With more data and more analysis, of course. We hoped to establish as precisely as possible how much heat these spacecraft really were shedding unevenly and whether the resulting force could explain the peculiar trajectories. For that, we needed two key sets of information: the “housekeeping data” engineers had used to keep tabs on spacecraft operation and the Doppler data, as much as we could find.

As luck would have it, most of the Pioneer 10 and 11 telemetry data had been saved and were available for study. Although there was no requirement that NASA properly archive these records, it turned out that systems engineer Larry Kellogg, a contractor and former Pioneer team member at NASA Ames Research Center, had been informally preserving all the Pioneer data he could get his hands on. Kellogg already had nearly all of the two probes’ master data records, binary data files that contained all the Pioneers’ science and housekeeping data.

Kellogg had taken care to copy those records, which in total took up just 40 gigabytes of space, from soon-to-be obsolete magneto-optical discs to a laptop hard drive. When we decided to work with the telemetry data in earnest, in 2005, one of us (Toth) had already been in touch with Kellogg, working on new software that could extract useful information from the master data records without the need for an old, decommissioned mainframe.

Procuring additional Doppler data that could help solve the mystery turned out to be a bit trickier. The JPL team had already collected all the radio-science data files that were easy to find and work with, but we knew that we needed more measurements. It took some time, but we were able to find additional files on the hard drives of JPL navigators’ computers and the archives of the National Space Science Data Center. We even found magnetic tapes stuffed in cardboard boxes under a staircase at JPL. Some of the files were in a rather sorry state, corrupted while they were converted from one storage format to another over the span of three decades.

To find these files and make them usable, we received financial support from the Planetary Society, a nonprofit space-advocacy organization. We are also grateful to physicist Craig Markwardt at NASA’s Goddard Space Flight Center, who helped recover corrupted files and get all the Doppler data into usable condition. In the end, we were able to double the amount of available Doppler measurements, increasing the span of orbital data to 23 years for Pioneer 10 and to more than 10 years for Pioneer 11.

At JPL, veteran navigators Neil Mottinger and Jordan Ellis analyzed the expanded Doppler observations. As with the original orbital analysis, they had to take into account all the information we had on anything that might affect the transmissions. They had to consider Earth’s spin, the locations of the receiving antennas, and both terrestrial and solar weather. They also had to factor in the changing communications strategies over the span of time—31 years for Pioneer 10 and 17 for Pioneer 11—during which the probes were capable of sending back signals that could be used for precision navigation. With their completed analysis in hand, we found that the Doppler data were still statistically consistent with a constant deceleration but that a value that decreased over time fitted the data even better.

If the effect diminished slightly over time, we reasoned, maybe the heat generated by the RTGs, which would have slowly decayed, was responsible for the anomaly after all.

To find out, we enlisted the help of a team of engineers at JPL led by Gary Kinsella, who had plenty of experience modeling the thermal behavior of spacecraft, to construct a detailed thermal model of Pioneer 10 and 11. In the absence of computer-aided-design files, which didn’t exist 40 years ago, Kinsella’s team had to comb through old blueprints and consult with the retired TRW engineers who built the probes in order to reconstruct their 3-D structure.

The team’s eventual geometric model divided the surface of each spacecraft into 3300 patches and incorporated documented thermal properties of the materials used to construct the probes. We used the model to show how heat flowed across and radiated off various parts, and with those results, we estimated the magnitude of the thermal recoil force at different times over the course of the Pioneer missions. After matching the model to the Pioneers’ temperature and electrical readings, we found that the spacecraft did experience a sizable thermal recoil force, corresponding to an excess of about 60 W even after 20 years in deep space. The magnitude of the force was still tiny by Earth standards—about the same as the backward push your car experiences in reaction to the photons spit out by its high-beam headlights. The team found that a good half of the force came from heat from the RTGs, which bounced off the back of the spacecraft antenna. The other half came from electrical heat from circuitry in the heart of the spacecraft. Most of that heat was radiated through louvers at the back of the probes, which weren’t as well insulated as the rest of their bodies, further contributing to the deceleration.

We also calculated the thermal recoil force on the spacecraft using just the Doppler data, by computing the force needed to match the probes’ trajectories. When we compared this independent estimate with the one derived from the spacecraft model, we found that the two values matched within 20 percent. Once uncertainties are taken into account, there is no statistically significant difference. Three decades after its discovery, we can now say there is no exotic cause for the Pioneer anomaly: The puzzling deceleration was produced by the asymmetric radiation of waste heat created onboard the spacecraft.

For some, this might seem like a disappointing conclusion to such a long-standing mystery, but accounting for the anomaly actually cements Pioneer 10 and 11’s place in history. These spacecraft accomplished something that isn’t likely to be repeated anytime soon: They completed a high-precision validation of Einstein’s theory of how gravity works, out to twice the distance between Pluto and the sun. No other spacecraft launched since could have been used for such a test (or would show such a subtle anomaly); later spacecraft were too reliant on thrusters, weren’t tracked continuously, or just didn’t travel far enough. Pioneer 10 and 11 remain our most precisely navigated deep-space probes to date and will probably hold that distinction for many years to come.

These spacecraft also underscore the value of data preservation. In the early days of the Pioneer missions, scientists and engineers often viewed the medium as more valuable than the data it contained. Many considered raw data to be worthless once “useful” scientific and technical information had been extracted. Nowadays data storage may be cheap, but we’re still in danger of suffering from shortsightedness when it comes to data custodianship. Every experiment needs a clear plan in place to ensure that a record of the original observations is still available and readable, even decades into the future. It may very well be the only way we’ll resolve the next confounding mystery.

About the Authors

Viktor T. Toth is a software engineer in Ottawa, Ont., Canada, and Slava G. Turyshev is a theoretical physicist at NASA’s Jet Propulsion Laboratory, in Pasadena, Calif. They each devoted years to solving the mystery of the Pioneer anomaly. Turyshev recalls many late nights spent reading navigational data from more than 400 magnetic tapes he’d found in moldy cardboard boxes beneath a staircase at JPL. “It took about as much time to clean the tape reader as it did to actually use it,” he says.

This article is for IEEE members only. Join IEEE to access our full archive.

Join the world’s largest professional organization devoted to engineering and applied sciences and get access to all of Spectrum’s articles, podcasts, and special reports. Learn more →

If you're already an IEEE member, please sign in to continue reading.

Membership includes:

  • Get unlimited access to IEEE Spectrum content
  • Follow your favorite topics to create a personalized feed of IEEE Spectrum content
  • Save Spectrum articles to read later
  • Network with other technology professionals
  • Establish a professional profile
  • Create a group to share and collaborate on projects
  • Discover IEEE events and activities
  • Join and participate in discussions