On 10 July 2025, the Earth finished one rotation in 1.36 milliseconds short of the usual 24 hours a record in a string of remarkably brief days in recent times. For most people, a millisecond is lost without notice. But for the world infrastructure supporting navigation, finance, communications, and science, these brief moments are seismic events, precipitating fundamental questions about how humans keep time and coordinate the planet’s most essential systems.
The effect is not a curiosity of astronomy. Since 2020, astronomers have measured a sustained speeding up of Earth’s rotation, defying the long-term trend of slowing down due to moon-induced tidal friction. As Dirk Piester of Germany’s national weather institute described, “We now have slightly shorter days than in the last 50 years.” The shortest days this year July 9, July 22, and the predicted August 5 are all likely to be less than 24 hours by more than a millisecond, with the position of the moon as far from the equator as possible magnifying the effect.
The causes behind these variations are intricate and, sometimes, slippery. Although the moon’s pull continues to be a commanding force, new research indicates Earth’s liquid core, atmospheric angular momentum, and even the transfer of mass from melting polar ice. According to Leonid Zotov, a premier expert on Earth rotation, “The cause of this acceleration is not explained. Most scientists believe it is something inside the Earth. Ocean and atmospheric models don’t explain this huge acceleration” recent changes in rotation.
These slight variations have disproportionate effects on worldwide timekeeping. The globe’s standard time, Coordinated Universal Time (UTC), is defined by an ensemble of more than 400 atomic clocks, each counting the hyperfine transition of cesium-133 atoms with accuracy to a few billionths of a second. UTC is not, however, merely atomic time. It is regularly tweaked to remain within 0.9 seconds of astronomical time, which is measured by Earth’s actual rotation a process that, since 1972, has necessitated the addition of 27 “leap seconds” leap second history.
Leap seconds, although theoretically pretty, have turned out to be a technical minefield. According to Patrizia Tavella, head of the time department at the International Bureau of Weights and Measures, “Leap seconds often cause failures and anomalies in computing systems.” The aviation sector, for instance, experienced flight scheduling disorder when various networks manage leap seconds differently. The 2012 leap second caused downtime for top websites and revealed weaknesses in financial and telecommunication systems leap second interruptions. As Douglas Arnold, principal technologist at Meinberg-USA, has observed, “In practice, problems arise because timing is distributed among a large number of devices in a network all designed by different people. So there are inconsistencies with when pending leap second events are announced, how long the information takes to propagate through the network, and what happens if there is a missing or extraneous leap second notification which is later corrected.”
The possibility of a “negative leap second”–subtracting a second from UTC to correct for the Earth spinning more quickly–has never actually been tried. Judah Levine, a physicist at the National Institute of Standards and Technology, warned, The primary concern about a negative leap second is that it has never happened before, and the software needed to implement it has never been tested. There are continuing problems with the insertion of positive leap seconds even after 50 years, and this increases the concerns about the errors and problems of a negative leap second. The risk is not hypothetical: in 2022, an industry “Plugfest” discovered that one device interpreted a negative leap second as a positive one, purely because its software could not cope with a second being taken away negative leap second software problems.
The wider timekeeping world is now at a choice point. In 2022, an international poll established a deadline for doing away with leap seconds by 2035, supported by both science agencies and technology giants. The reason is obvious: as atomic clocks and synchronization networks continue to improve, the disruptive price of leap seconds outweighs the value of maintaining civil time tied to the Earth’s irregular rotation. As David Gozzard, an experimental physicist, put it, “Atomic clocks and our computer networks are the new, far superior form of time measurement, but we’re forcing them to keep in sync with this older form of measurement.”
Yet, the transition is fraught with trade-offs. Abandoning leap seconds will allow UTC to drift from solar time, gradually decoupling noon from the sun’s zenith. For all but the most applications, such a drift will be imperceptible for centuries, but for astronomy, navigation, and geodesy, the difference is still crucial. A few have suggested alternative time bases for technical systems e.g., International Atomic Time (TAI) or GPS time and ones that don’t include leap seconds and are already utilized for critical synchronization in GNSS and network operations atomic time standards.
At the same time, the physics of timekeeping is getting better. Optical clocks that measure the vibrations of visible light instead of microwaves now outperform cesium standards in terms of accuracy and stability. These new-generation clocks have the potential to redefine the SI second and further refine the accuracy of global time dissemination networks optical clock advances.
As Earth’s rotation plows on in its erratic waltz, the planet’s timekeepers encounter a paradox: the more accurately humanity can keep time, the more evident Earth’s own anomalies grow. The move to end leap seconds, or risk the unprecedented action of a negative leap second, is less a technical aside than a deep accounting with the boundaries of synchrony in a world where even the duration of a day is no longer fixed.
