The most dramatic-looking feature of the Pioneer spacecraft was a 2.74 m diameter parabolic dish used to stay in communications with earth. One of its tasks was to take a 2.1-GHz S-band carrier signal transmitted from earth and coherently up-convert it by a factor of 240/210 to about 2.29 GHz and retransmit it back.
Since the received frequency stability was as accurate as the earth-station's frequency stability, small frequency shifts on the order of 1-mHz could be detected. This is a frequency shift of a part in 1 trillion over a few hours.
Precision measurements of frequency shifts received on earth could be interpreted as the Doppler shift in the signal. This enabled precision measurement of the spacecraft's line of sight velocity, which could be translated to position and acceleration information.
By 1980, an anomaly had been detected in its motion. The unexplained frequency shift was only a few tens of milliHertz, but it was far above the noise. The spacecraft was slowing down more than expected. The deceleration was about 0.09 nano gees. This is about the same deceleration a car would have from the photon pressure from its high beams.
It could not be accounted for by any known influence, such as the gravitational effects of the sun or planets, the solar wind, cosmic ray fluxes, the precession of the poles of the earth, even residual fuel leakage from the spacecraft. This slight deceleration became known as the Pioneer Anomaly.
It appeared the acceleration was constant and not changing over time. This seemed to confirm the early estimate that the influence of thermal radiation from the radioisotope thermoelectric generators (RTG) was not contributing significantly to the deceleration. If it was a thermal radiation effect, the 88-year half-life of the Plutonium 238 should have contributed to a change in the deceleration as the thermal radiation decreased.
In 1994, out of the blue, Anderson got an email from Michael Nieto, a Los Alamos cosmologist. He was interested in MOND, MOdified Newtonian Dynamics, theories of gravity. Coincidentally, Nieto thought he could use precision measurements of the Pioneer satellite as a test for non-Newtonian gravity effects. He expected anomalous accelerations on the order of the expansion rate slow down of the universe, about 0.07 nano Gees. Here was one possible explanation. Wow! Could this agreement just be a coincidence?
At around the same time, Slava Turyshev joined Anderson to work on the Pioneer Anomaly reviewing all possible spacecraft-related causes for the anomaly.
In 1998, what was known of the Pioneer Anomaly was published to stimulate the scientific community to come up with other explanations.
Almost 1,000 papers were published with explanations for the Pioneer Anomaly, including dark energy, dark matter, string theory, violation of general relativity, extra dimensions, and even variations in the speed of light. The chart below shows the publication rate in papers just posted on arXiv over the years.
But Anderson's team was not willing to let go of possible conventional explanations. They managed to find more historical Doppler data, some on old magnetic tape media, stuffed in moldy cardboard boxes under a staircase at JPL. With the new data, they increased the accuracy and time span of Pioneer 10 measurements to more than 23 years. In the new data, they saw a change in the deceleration. It was decreasing over time.