There are a few things in this cosmos that are truly universal. The fundamental laws that govern reality apply everywhere and at all times. The fundamental constants that determine the masses, charges, and interaction strengths between particles are universal as well. Today, there are four known fundamental forces:
- electromagnetism,
- gravitation,
- the strong nuclear force,
- and the weak nuclear force,
plus whatever “dark energy” ultimately turns out to be. Electromagnetism is, in many ways, the best-measured of all the fundamental interactions, and its strength is known to better than 1-part-in-a-billion from precision laboratory experiments.
But gravitation, despite being the first fundamental force ever discovered, remains extremely poorly known: to only 1-part-in-a-few-thousand, where the uncertainty shows up in the fourth significant digit. Part of the reason is due to its incredible weakness; the gravitational attraction between two electrons is more than forty orders of magnitude smaller than the electrostatic repulsion between two electrons. But part of the reason is that even with our most precise and accurate measurements that lead to a determination of the gravitational constant, those various teams and methods haven’t converged on a single answer.
In April of 2026, a new measurement of the gravitational constant was announced: the culmination of a decade of efforts at NIST, the National Institute of Standards and Technology. The mystery over just what this constant is, G or “big G,” has now deepened even further. Here’s the story behind what we’ve learned.
As Earth orbits around the Sun, it rotates on its axis. Although the speed of Earth’s motion is slow compared to the speed of light, moving at just 30 km/s in orbit, or 0.01% the speed of light, it’s much faster than the rotational speed, which maxes out at the equator at a much more modest 1670 km/hr (0.47 km/s). Although it’s easy to compute the laws and equations of motions for the Earth-Sun system, for example, this doesn’t provide sufficient information to arrive at a value for G, the universal gravitational constant.
Credit: Sophie DesRosiers/Montréal Space for Life
The quest to understand the Universe scientifically, in many ways, began with gravity. In ancient times, every human was greeted with a dazzling night sky, filled with thousands of glittering points of light; every night, most of those points would appear in the same relative position, but the Moon and planets would wander on a night-to-night basis. Similarly, all objects on Earth accelerate downward, toward the center of our planet, and are only prevented from falling through it by the force of the Earth pushing back up on them. While many early scientists made models to explain planetary motion and worked to measure the gravitational acceleration at Earth’s surface, what we call g or “little g” in physics, it wasn’t until Isaac Newton came along that we realized that the same underlying force and phenomenon — gravitation — was responsible for both.
Newton’s law of universal gravitation represented that remarkable connection: between the terrestrial and the celestial. Newton calculated the rate at which the Moon “fell” or accelerated toward the Earth, and he was the first to place that acceleration on the same footing that explained accelerations due to gravity at Earth’s surface. He recognized that the orbits of Jupiter’s moons were determined by Jupiter’s gravity, and the motions of the planets around the Sun were determined by the Sun’s gravity. In the late 17th century, he published his law of universal gravitation, which explained Kepler’s laws of planetary motion and the orbits of periodic comets alike in one fell swoop.
Newton’s law of universal gravitation (left) and Coulomb’s law for electrostatics (right) have almost identical forms, but the fundamental difference of one type vs. two types of charge open up a world of new possibilities for electromagnetism. In both instances, however, only one force-carrying particle, the graviton or the photon, respectively, is required.
Credits: Dennis Nilsson/RJB1, Wikimedia Commons
But there was a problem: the universal gravitational constant that Newton had proposed, G or “big G” in physics terminology, wasn’t directly measurable from tracking the orbits or accelerations of objects that influenced them. When a planet orbits the Sun, it accelerates around it due to a combination of the universal gravitational constant (G) and the Sun’s mass (MSun.). When the moons of Jupiter orbit Jupiter, they are accelerated by a combination of the universal gravitational constant (G) and Jupiter’s mass (MJup). When objects, whether the Moon, a satellite, or something right here on Earth’s surface, are accelerated by the Earth’s gravity, it’s due to a combination of the universal gravitational constant (G) and Earth’s mass (ME).
If we want to know what the universal gravitational constant is, we need some way to disentangle it from the uncertain mass of the object that’s causing the acceleration. This is practically impossible for something with an enormous mass like a planet or star, as there’s no known way to disentangle the gravitational constant (G) from a mass (M) from orbital mechanics alone. That’s why it’s so profound that, in the late 1700s, Henry Cavendish — building on the earlier work of his mentor John Michell — implemented a method to measure G independently of any mass in the Universe, using a device known as a torsion balance.
This schematic of the Eötvös torsion balance, as modified by Jakosky in 1940, shows how the equivalence principle can be tested in two dimensions: in both the horizontal and vertical directions, by suspending one mass below the other in a torsion balance setup. This was an outgrowth of the original torsion balance: designed by John Michell and completed by Henry Cavendish.
Credit: R.J. Howarth, Gravity Surveying in Early Geophysics, 2007
The way a torsion balance works is as follows.
- First, you construct a stable, horizontal beam that’s elevated high off of the surface of the Earth.
- You build a closed box around that beam (and everything below) to protect the environment from air flow and air currents, as well as other ambient effects.
- Next, you hang two small, spherical masses that are supported by a fine (torsion) wire in the center.
- Then, you’d attach two much larger masses that were attached to a separate, independently fixed suspection.
- And finally, you’d let the experiment run: allowing the smaller masses to be attracted to the larger ones, causing the beam to rotate (slightly) and the torsion wire connecting the two small masses to deflect (slightly).
What is it that causes the rotation of the beam and the deflection of the torsion wire? It’s gravity, of course. But in this case, it isn’t the gravity of Earth, or the Sun, or any planet at all that causes these measurable effects, but the gravity between the masses used in the experiment: something that’s measurable directly, and that’s independent of a (highly uncertain) planetary, lunar, or stellar mass.
It was Michell who initiated this idea, designing the apparatus in 1783 and beginning work on building it thereafter. However, it remained unfinished in 1793: the year Michell died. After passing through the hands of Francis Wollaston, the device and its design fell into the hands of Henry Cavendish: already well-known for having discovered and isolated what we now know as hydrogen: the first atom in the periodic table of elements.
These two figures show a face-on view (top) and top-down view (bottom) of the Cavendish experiment, as illustrated by Henry Cavendish himself back in 1798. The large and small masses, the torsion balance, and the horizontal wooden beam supporting the experimental apparatus can all be easily identified here.
Credit: Henry Cavendish/public domain
Cavendish wound up rebuilding the entire apparatus from scratch, although he adhered closely to Michell’s original design. The large masses that he used were enormous and dense: 0.3 meters in diameter (12 inches) apiece and made of lead, coming in at 158 kg (348 lbs) apiece, and he completed the apparatus’s construction in 1797. The torsion balance rod, with smaller masses attached to it, did indeed deflect as predicted. The deflection was measured to be by an amount that depended on the stiffness of the suspending wire: 0.16 inches (0.41 cm) when a normal suspending wire was used and a mere 0.03 inches (0.076 cm) with a stiffer suspending wire.
The research was completed in 1798, and Cavendish went ahead and then published his results. From the angle of deflection of the torsion rod, combined with a knowledge of how the torque on the wire behaves from a physical perspective, one could determine the magnitude of the forces between the various masses in the problem. Even though the mass of the Earth was not well-known, the magnitude of the gravitational force on each of the masses from the Earth (which relies on that combination, GME) was well-understood.
With just a little bit of physics calculations, Cavendish could use the ratios of the force exerted by the Earth on the masses to the forces exerted on the masses by one another to calculate:
- the value of the gravitational constant, G,
- the value of the mass of the Earth, ME,
- and, because we knew the size of the Earth (and hence, its volume), the value of the density of the Earth.
For that last figure, Cavendish obtained a value of 5.480±0.038 grams-per-cubic centimeter: double the density of crustal rocks found at Earth’s surface, and consistent with the modern value of 5.514 grams-per-cubic centimeter.
This diagram shows a torsion balance experiment similar to the one used by Cavendish (and later scientists) in attempts to measure the gravitational constant directly. The large masses (M) attract the small movable masses (m), which in turn cause the torsion wire to deflect from their equilibrium position in the absence of those masses.
Credit: Chris Burks (Chetvorno)/Wikimedia Commons
That’s the origin story for how we first obtained the measurements necessary to determine G, the universal gravitational constant. Using Cavendish’s values, and expressed in modern units of meters-cubed-per-kilogram-per-second-squared, we get a value for G = 6.74 × 10-11, which differs from the best modern value only about 1%. This means that the first determination of the universal gravitational constant, G, was good enough to have just a one-part-in-100 uncertainty to it.
You’d probably imagine, given all the advances in science and technology that have occurred in the 228 years that have passed since Cavendish published his first results, that we’d now be doing much, much better in the quest to measure G. Indeed, it did improve:
- John Henry Poynting, in 1891, improved the error to just 0.2% (1-part-in-500),
- Paul Heyl, in the 1930s and 1940s, improved it to 0.1% (1-part-in-1000),
- and then various teams, who owe a great debt to Barry Taylor in the 1980s and 1990s, published values for G with uncertainties as low as 0.012% (1-part-in-8000).
The torsion balance remained an important part of this story, but was joined by other methods. One could place a large, nearby mass near a hanging pendulum, measuring the deflection of the pendulum induced by the gravity of the mass. One could use beam-balance scales instead of a torsion balance, where the scale would be dually sensitive to masses that rest upon it as well as the external influence of gravitational sources. Or one could leverage atom interferometry: a technique that only returned the first successful measurements of G in 2014.
This graph shows the evolution in our best measurements of G, the gravitational constant, as a function of time. The sizes of the error bars have been variously small and large, and many of the 1980s and 1990s “consensus” measurements (that had small error bars and low values) are omitted here. A new experiment in 1998, with large error bars as shown, helped bring about our modern measurements of G. Red points are recommended values based on meta-studies; blue points are torsion balance experiment results; green points represent values from other methods, including pendula, beam balances, and atom interferometry.
Credit: Dbachmann/Wikimedia Commons
While many thought that we were honing in on the true value of G, a problem swiftly emerged. Different groups, whether using the same method or different methods, were getting wildly different values. Measurements of G were yielding results as low as G = 6.671 × 10-11, while others were as high as G = 6.676 × 10-11. Around the early 2000s, we were forced to admit that teams claiming 0.012% uncertainties were facing much larger, perhaps ill-documented or even unidentified errors, and that we really only were certain about the value of G to about a 0.05% (1-part-in-2000) margin.
Then, in 2018, a group from China built and created the most accurate and precise torsion balance of all-time, obtaining two excellent measurements for G: G = 6.674484× 10-11 and G = 6.674184× 10-11, with claimed uncertainties of only 0.001%, but which differ from each other at five times that amount. Moreover, other groups frequently obtained smaller results. More than 200 years after Cavendish, we still couldn’t even claim to definitively know what the fourth digit in the gravitational constant was.
The two methods of experimental setup published at the end of August, 2018, in Nature, which yielded the most precise (claimed) measurements of G to date, but whose values are somewhat higher than the world average of all measurements and with suspected unquantified systematic uncertainties.
Credit: Q. Liu et al., Nature, 2018
That’s part of what led NIST scientist Stephan Schlamminger to embark on a 10-year quest to measure the gravitational constant, G, to greater precision and with greater accuracy and lower uncertainty than ever before. As first reported by Ron Cowen at the National Institute of Standards and Technology (NIST), those results were finally released in April of 2026.
There’s always a risk, when you conduct science yourself, that you’re going to be biased — even if you try to fight against it, intellectually — by your expectations for the outcome, which is usually driven by prior results. That’s why Schlamminger, this time, asked his colleague Patrick Abbott (who wasn’t part of the experimental team) to scramble the data: so that even Schlamminger himself wouldn’t know what the true masses in the experimental setup were. Abbott sealed the code to uncovering the true values of the masses in an envelope, and Schlamminger performed the experiment without ever opening it.
At last, in 2026, the research was published, and the big number was finally revealed. Their measurement for G was G = 6.67387× 10-11, with a reported uncertainty of just 0.0057%.
However, this number, again, does not align with either similar or complementary experiments. Schlamminger’s experiment was an attempt to reproduce an earlier French experiment, but his team got a value that was a full 0.025% lower: about five times larger than the reported uncertainty. It’s approximately 0.01% lower than the 2018 results from China. And it truly, in many ways, only deepens the mystery over the “true value” of G.
This graph shows many different measurements and determinations of G, the gravitational constant, through a variety of techniques since the early 1980s. Values have varied significantly, and different experimental setups and teams have failed to converge on a single, universal value, even though we can be confident that G hasn’t changed in billions of years.
Credit: S. Schlamminger et al., Metrologia, 2026
Nevertheless, the new experimental results are important, and even groundbreaking, in a variety of ways.
- The research team led by Schlamminger performed a completely blind analysis, randomizing the values of the masses used to prevent experimenter bias. This should become part of standard best practices in all future analyses.
- The results do not corroborate either the current (CODATA) recommendations or the earlier (BIPM) values that this particular experiment was designed to emulate. Importantly, if you include all uncertainties associated with the current recommendations and this experiment, the values are not statistically significantly different.
- Even though the team did not identify a single main cause behind the discrepancy with earlier (BIPM) results using the same method, they did identify an array of systematic effects that were previously overlooked: the tri-lobed shapes of the test masses, the gas-pressure forces that produce a torque during the motion of the source mass, the non-linearity of the autocollimator, and what they call a “dark uncertainty” implied by a discrepancy between individual, successive measurements.
They also considered that simply by underestimating a distance by as little as one micron, it would affect the overall measurements of G by over a dozen parts-per-million, while thermal torques or a residual pressure in the vacuum could affect measurements by nearly as large an amount.
This animation shows the setup at NIST for measuring the strength of gravity: an advancement over the original torsion balance method of Cavendish in several ways. The additional test and source masses, plus the laser used to measure the deflection due to gravitation, enables many variants to be conducted, which hopefully will someday lead to a precise, accurate, and definitive determination of G.
Credit: S. Kelley/NIST
This is actually very important to the advancement of science. As the authors write in their paper:
“Finally, we note that our total uncertainty is larger than in the previous determination. This increase does not reflect a degradation in measurement performance, but rather a more complete accounting of systematic limits. By explicitly including the ‘dark uncertainty’ driven by the scatter among our individual datasets, we believe this result provides a more robust estimate of the true accuracy achievable with this apparatus.”
Sure, it would have been nice to get an answer that was wholly consistent with previous answers. It would have been wonderful to get a more precise, more accurate answer that pinned down the value of G once and for all. And it would’ve been nice to identify all the sources of uncertainty in current and past experiments that account for the differences obtained between them over the past 40+ years. But that’s very, very rarely how science works in practice. Instead, this latest effort represents the incremental progress essential to experimental science, where errors and uncertainties are identified and quantified, and future generations of scientists can build upon it. After all, among all the disciplines in the world, science is the one most devoted to creating our best approximation of truthful reality. In that sense, every measurement is important, with the absolute strength of the gravitational force representing one of the most fundamental pursuits of all.