Eötvös’s measurements stood as the benchmark for decades, offering a firm basis for Einstein’s assumptions. Then in the early 1960s astronomer Robert H. (Bob) Dicke of Princeton, along with colleagues G. Roll and R. Krotkov, suggested a clever way of substantially improving on Eötvös’s method. Realizing that the Sun exerts a periodic pull on terrestrial objects—due to Earth’s 24-hour rotation—they measured the accelerations of various objects with respect to the Sun rather than Earth. The device they used was a triangular array of weights: two made of aluminum and one made of gold. An electrical system served to keep the set balanced. If any of the weights felt an extra tug and the device started to tilt ever so slightly, an electrical signal would immediately rectify it. The amount of this signal was carefully recorded.
Now suppose the equivalence principle was false and acceleration depended on mass. Then aluminum would react slightly differently than gold to the Sun’s pull. As Earth turned around on its axis, the torsion balance would try to tilt slightly in different directions. The electrical system would thereby need to exert a periodic correction—with an unmistakable 24-hour cycle. Dicke and his co-workers found no such cycle. Within a difference of one part in 100 billion, they confirmed that aluminum and gold accelerate at the same rate under gravity.
Refining Eötvös’s concept even further, in the 1990s a group of researchers led by Eric Adelberger of the University of Washington constructed several torsion balances with even greater sensitivity. They designed each balance to test particular features of gravity on a variety of scales. Rather than just looking at the effects of Earth and the Sun, they fashioned their instruments to measure gravitational influences as close as the fly’s wings and as far away as the center of the Milky Way. To honor both Eötvös and their university, they named their collaboration the Eöt-Wash group—pointing out that “vös” in Hungarian is pronounced somewhat like “Wash” in English.