What Is a Leap Second?
A leap second is a one-second adjustment that is occasionally added to Coordinated Universal Time (UTC), the global time standard your phone, servers, and satellites all sync to. When one is inserted, the last minute of the day contains 61 seconds instead of 60 — the clock ticks through 23:59:60 before rolling over to 00:00:00.
Leap seconds sound trivial. They are not. That extra second has crashed major websites, grounded flights, and triggered a decades-long argument among physicists, astronomers, and software engineers. In 2022 the world's timekeeping authorities finally voted to get rid of them. Here is what a leap second actually is, why it exists, and why it's on its way out.
Why Leap Seconds Exist
The problem comes down to two different ways of measuring time that don't quite agree.
Atomic time is perfectly steady. Since 1967, the second has been defined by the vibrations of cesium-133 atoms — 9,192,631,770 of them per second. A network of atomic clocks around the world keeps International Atomic Time (TAI), and it is astonishingly regular: it neither speeds up nor slows down.
Earth's rotation is not steady. The length of a day, as measured by the planet's actual spin, is slightly irregular. Friction from tides, movement of Earth's molten core, and mass shifts like melting ice sheets all nudge the rotation rate. Over the long run, the planet is generally slowing down, so an astronomical day drifts a tiny bit longer than 86,400 atomic seconds. This rotation-based time is called UT1.
Left alone, UTC (built on steady atomic seconds) and UT1 (tied to the actual Sun in the sky) would slowly drift apart. To keep them aligned, the rule is simple: UTC must never differ from UT1 by more than 0.9 seconds. Whenever the gap approaches that limit, a leap second is inserted into UTC to let atomic time "wait" for the slower-spinning Earth to catch up.
So a leap second is essentially a manual correction that keeps our precise atomic clocks in step with the wobbling planet we actually live on.
A Short History of Leap Seconds
Leap seconds were introduced in 1972, when UTC was redefined to be based on atomic time while staying tied to Earth's rotation.
Since then, 27 leap seconds have been added through 2016. Every single one so far has been positive — a second added, never removed — because Earth's rotation has generally lagged behind atomic time over that period. The most recent leap second was inserted on 31 December 2016, at 23:59:60 UTC.
There was a notable gap of nearly seven years after that, because Earth's rotation actually sped up slightly in recent years — enough that, for the first time, scientists began seriously discussing the possibility of a negative leap second (removing a second rather than adding one), something that has never happened and that many engineers dread even more than the usual kind.
Who Decides When One Happens?
Leap seconds are not scheduled on a fixed calendar. They are announced by the International Earth Rotation and Reference Systems Service (IERS), headquartered near Paris, which continuously monitors the planet's rotation.
When the UT1–UTC gap is heading toward 0.9 seconds, the IERS issues a bulletin — typically about six months in advance — declaring that a leap second will be inserted. By convention, insertions happen at the end of either June 30 or December 31 (UTC).
That six-month lead time is the crux of the engineering problem: the schedule is irregular and unpredictable. You cannot write a formula that tells you when the next leap second will occur, the way you can for leap years. You have to wait for the IERS to announce it, then make sure every clock in your system knows.
Why Engineers Hate Leap Seconds
For most people, an extra second is unnoticeable. For computers that expect every minute to have exactly 60 seconds, a 61-second minute is a landmine.
The most famous incident came on 30 June 2012, when a positive leap second was inserted. The moment it hit, a range of systems fell over. Reddit went down, along with other large sites. The bug traced back to how the Linux kernel handled the clock stepping backward for that instant — it triggered a runaway condition that pegged servers' CPUs. Airline reservation systems built on affected software also reported disruptions, delaying check-ins. Similar trouble had appeared in earlier leap seconds too.
The root issues are structural:
- The event is rare, so leap-second-handling code paths are almost never exercised and rarely tested under real conditions.
- Time appears to stand still or repeat, and a lot of software quietly assumes time only ever moves forward, monotonically.
- The schedule is unpredictable, so systems must fetch and trust an external announcement rather than compute the adjustment themselves.
Because a modern outage can cascade across financial markets, telecom networks, and cloud platforms, a single mishandled second can be genuinely expensive.
Google's "Leap Smear" Workaround
Rather than apply the leap second in one abrupt jump, Google pioneered an approach called the leap smear. Instead of inserting a full second at midnight, Google's servers spread the correction across a long window — typically the ~24 hours surrounding the event — by making each second imperceptibly longer (or shorter).
During a smear, no clock ever shows 23:59:60 and no clock ever steps backward; time just runs very slightly "slow" for a day until the extra second has been quietly absorbed. Software sees a smooth, monotonic clock and never hits the dangerous edge case. Amazon, Meta, and others adopted similar smearing strategies, and it has become the de facto survival technique for large-scale systems.
The catch: a smeared clock deliberately disagrees with official UTC for hours at a time — fine for most services, but a headache when you need to reconcile it with unsmeared systems. It's a workaround for a problem the industry increasingly felt shouldn't exist at all.
If you work with epoch time, note that a Unix timestamp does not count leap seconds at all — Unix time simply pretends every day has exactly 86,400 seconds, which is part of why the real-world adjustment causes so much friction.
The 2035 Decision: Leap Seconds Abolished
The debate came to a head at the 2022 General Conference on Weights and Measures (CGPM) — the international body that governs the metric system and the definition of the second.
Delegates passed a landmark resolution: the practice of adding leap seconds will be stopped by 2035. From that point, UTC will be allowed to drift away from UT1 by more than the current 0.9-second limit, with the tolerance to be renegotiated later — likely allowing the gap to grow for decades, perhaps to a minute or more, before any correction is even considered.
In practical terms, this means UTC becomes a smooth, continuous timescale with no more surprise 61-second minutes. Astronomers who genuinely need Earth-rotation time can still use UT1 directly; everyone running computers gets a clock they can finally trust to behave. The change trades a tiny, slowly accumulating disagreement with the Sun for the elimination of a recurring source of global technical risk.
Frequently Asked Questions
Has a leap second ever been removed instead of added?
No. All 27 leap seconds inserted since 1972 have been positive — a second added. A negative leap second (skipping from 23:59:58 to 00:00:00) has never occurred, though Earth's recent faster spin briefly made one a real possibility before the 2035 decision made the whole question moot.
Do leap seconds affect time zones?
Not directly. Time zones are fixed offsets from UTC managed through databases like the IANA time zone database. A leap second adjusts UTC itself, so every zone shifts by the same instant simultaneously — the offsets between zones never change.
Will my computer clock change in 2035?
No visible change for users. After 2035, clocks simply stop receiving the occasional leap second. UTC will slowly drift from astronomical time over many years, but that drift is far too small to notice in daily life.
What is the difference between UTC, TAI, and UT1?
TAI is pure atomic time, perfectly steady. UT1 is time based on Earth's actual rotation. UTC is the everyday standard: it runs on atomic seconds like TAI but was kept aligned to UT1 using leap seconds. As of 2016, UTC is exactly 37 seconds behind TAI.
Why not just let the clock and the Sun drift apart?
That is essentially what the 2035 decision does. The disagreement grows so slowly — on the order of a minute per century — that it will take generations to matter, and by then a one-time correction can be planned deliberately rather than patched every few years.
Curious how machines actually count time beneath all of this? Leap seconds vanish entirely in epoch time — try our Unix timestamp converter to see exactly how computers track the seconds ticking by.