1. What a Removed Second Would Look Like on a Clock
Watch a digital clock near midnight: 23:59:56, then 57, then 58, then 59, and then 00:00:00. A positive leap second adds an extra 23:59:60 after the 59. A negative one does the reverse. The clock would show 23:59:58 and then jump straight to 00:00:00, and the "59" would never appear. No time is taken out of the day; UTC simply skips the label 23:59:59.
The correction is applied to Coordinated Universal Time, or UTC, the time scale that civil clocks around the world are set from. It can only be made at the end of June or the end of December[1]. The U.S. National Institute of Standards and Technology (NIST) says the rules have always included "a provision for negative leap seconds if it becomes necessary due to changes in Earth's rotation"[1]. The provision exists, but it has never been used.
2. Why the World's Clocks Need Corrections at All
There are two ways to keep time. Atomic clocks tick at an extremely steady rate. The Earth turning once is less steady, and the time it gives, called UT1, drifts away from the atomic clocks. UTC is the compromise: it ticks like an atomic clock, but a whole second is added or removed as needed to keep it within 0.9 seconds of UT1[1][3]. A group called the International Earth Rotation and Reference Systems Service (IERS) measures the Earth's rotation and publishes that difference[1].
Whether a second has to be added or removed depends on the direction of the drift. NIST explains that, averaged over long periods, the Earth's rotation is gradually slowing, and that this is why leap seconds have so far all been additions: a slower Earth makes the day a little longer than the atomic clocks expect[1]. A removal would be needed only if the Earth were turning faster than the clocks, so that the Earth's time ran ahead.
3. A Faster Spin Pushes Toward Removing a Second, and Melting Ice Pushes Back
In March 2024 the geophysicist Duncan Agnew of the Scripps Institution of Oceanography published a paper in the journal Nature on this question. The university's news story about it says the motion of the Earth's liquid core, a mass of molten iron inside the solid planet, has been working to speed up the rotation of the solid Earth. Agnew suggested a negative leap second might be needed some time in 2028[2]. That is one researcher's estimate, not a decision, and rotation is hard to predict years ahead.
Here is the surprising part. When ice near the poles melts, the water spreads out toward the equator. The story describes this as a transfer of mass from the poles toward the equator, which slows the Earth's rotation[2]. A simple way to picture it is a spinning skater who stretches their arms out: they slow down. By that estimate, without the slowing from melting ice, the first negative leap second would have been expected three years earlier[2]. So global warming, in this one respect, is giving the world's timekeepers a few extra years.
This does not make melting ice a good thing. Melting ice is the cause of many other, much larger problems. It is simply a surprising example of how something happening on the whole planet shows up, in a tiny way, on the clocks.
4. Why Removing a Second Worries the People Who Run Computers
Most people would not notice a missing second in everyday life. To a network of computers it can. If one system adds or removes a second and another does not, the two disagree about the time. The UC San Diego story gives an everyday example: an online shop and your bank could end up with different clocks, so one click on "order" would look like two actions at two different times[2].
Adding a second has been done many times, and software has had practice. Removing one never has been. The October 2026 resolution of the General Conference on Weights and Measures (CGPM), the international body of member countries that oversees measurement, says a negative leap second has never been applied and "is considered to pose a high risk of causing anomalies and disruption to critical infrastructures"[5]. An earlier resolution, in 2022, had already warned that the jumps leap seconds create could risk "serious malfunctions in critical digital infrastructure," naming satellite navigation, telecommunications and energy transmission[3].
5. The Clock Keepers Voted to Stop Using Leap Seconds
In 2022 the CGPM decided that the allowed gap between UT1 and UTC, now 0.9 seconds, would be made larger in or before 2035, so that UTC could stay continuous for at least a century[3]. At its 28th meeting in October 2026 it went further. Resolution 5 says that "continuous UTC will become effective on 20 May 2027," and that the maximum difference will be 3,600 seconds, which is one hour[5]. The aim is that UTC would no longer have leap seconds put in or taken out.
The resolution gives reasons. A workshop of timekeeping experts in March 2025 estimated that the probability of a negative leap second "will increase rapidly in the near future," reaching about 30 percent by 2035. It adds that preparing for one is expected to cost roughly as much as the preparations for the Year 2000 computer bug[5]. (This is the resolution's own summary of the workshop's work; I did not read the workshop report.)
So what has actually happened lately? The IERS publishes a notice twice a year that says whether to add a leap second. Its Bulletin C 72, dated July 6, 2026, says no leap second will be added at the end of December 2026. UTC has stayed 37 seconds behind International Atomic Time since January 1, 2017[4]. Forecast, decision and rules are three different things. Researchers' forecasts say when the Earth may need a removal, the IERS notices say what happens at each half-year, and the CGPM decides what rules govern all of it.
6. Try Skipping a Second Yourself
Write out the clock numbers from 23:59:56 on paper: 56, 57, 58, 59, 00:00:00. Then cross out the 59 and write 00:00:00 right after the 58. Now read the sequence aloud, one number per second, imagining that the clock jumps from 23:59:58 to 00:00:00, and notice how short the skipped moment is.
If you want to dig further, the BIPM resolutions and the NIST page in the list below are official sources, though some parts use technical language, and each is short enough to read in one sitting. When you see a headline such as "the world will lose a second in 2028," try asking which of the three it is describing: a forecast, a decision, or a rule.
Sources
This article is a personal summary based on the public sources listed below. Each note says how far I could read. Web pages were read through a summarizing fetch tool, so the quoted lines are as that tool returned them; the pages were not read in full by hand.
- NIST, "Leap Seconds FAQs." https://www.nist.gov/pml/time-and-frequency-division/leap-seconds-faqs (Source for: all leap seconds so far being additions, the provision for negative ones, the 0.9 second limit, June/December timing, the IERS role, and the long-term slowing. Read the first part of the page; the page text was truncated by the tool.)
- UC San Diego Today, "Global Warming Is Influencing Global Timekeeping," about Duncan Agnew's paper in Nature, March 27, 2024. https://today.ucsd.edu/story/global-warming-is-influencing-global-timekeeping (Source for: the liquid core, 2028, melting ice slowing the spin, and the three-year delay. A university news story, not the paper itself. The skater picture is my own illustration, not from the source.)
- BIPM, Resolution 4 of the 27th CGPM (2022), "On the use and future development of UTC." https://www.bipm.org/en/cgpm-2022/resolution-4 (Source for: the 2022 decision to enlarge the UT1-UTC maximum by 2035, and the warning about discontinuities. Primary document; read the operative text.)
- IERS, Bulletin C 72, July 6, 2026. https://hpiers.obspm.fr/iers/bul/bulc/bulletinc.dat (Source for: no leap second at the end of December 2026, and UTC-TAI of -37 seconds since January 1, 2017. Primary notice; this address always shows the latest bulletin, so it will change over time.)
- BIPM, Resolution 5 of the 28th CGPM (2026), "On the technical actions needed to ensure the continuity of UTC." https://www.bipm.org/en/cgpm-2026/resolution-5 (Source for: continuous UTC from May 20, 2027, the one-hour maximum, the 30 percent by 2035 estimate, and the risk statements. Primary document; the page was truncated by the tool, so I read the operative decisions and reasons it returned.)
Update history
- First published.