Most days contain 86,400 seconds. On rare occasions, the world's official timekeepers add one more. That leap second exists because two excellent ways of measuring time do not stay perfectly aligned: atomic clocks count uniform physical transitions, while Earth rotates at a speed that changes slightly.
The adjustment is tiny for a person. For computers processing millions of timestamped events, an unexpected 23:59:60 can be a serious engineering problem.
Atomic time is extremely regular
The SI second is defined using a transition in the caesium-133 atom. Laboratories around the world operate atomic clocks and combine their measurements into International Atomic Time, known as TAI.
Atomic time does not speed up when Earth rotates faster or slow down when it rotates more slowly. It provides the stable interval needed by telecommunications, navigation, finance and science.
Earth's day is not perfectly constant
Astronomical time tracks Earth's orientation relative to the Sun and distant space. Tides, the atmosphere, oceans and motion within the planet all influence rotation.
Over very long periods, tidal friction generally lengthens the day. Over shorter periods, rotation can vary in either direction. A civil clock based only on uniform atomic seconds would therefore drift gradually away from the position of the Sun in the sky.
What UTC does
Coordinated Universal Time, or UTC, is the basis of civil time zones. It uses atomic seconds but is occasionally adjusted so it remains close to UT1, a measure derived from Earth's rotation.
The International Earth Rotation and Reference Systems Service monitors the difference. Under the current system, a leap second can be announced when needed to keep UTC within the specified tolerance of UT1.
Time zones such as IST, GMT offsets and daylight-saving schedules are expressed relative to UTC.
What happens during a positive leap second?
A normal clock near midnight progresses from 23:59:59 to 00:00:00. During a positive leap second, the sequence can be represented as 23:59:59, 23:59:60 and then 00:00:00.
The extra second is introduced simultaneously worldwide, though the local clock time depends on the time zone. International timekeeping authorities announce it in advance.
A negative leap second—skipping a second—has been possible in principle but has not been used historically.
Why computers dislike leap seconds
Many software systems assume every minute has exactly 60 seconds and that timestamps always move forward in a familiar sequence. The label 23:59:60 may fail validation, cause repeated timestamps or confuse ordering.
Distributed databases, trading platforms and logging systems need consistent answers about which event occurred first. A one-second discontinuity can expose bugs that ordinary testing never encounters.
Different operating systems and services can also apply the adjustment differently, briefly disagreeing about the current time.
What is leap smearing?
Some technology operators use a leap smear. Instead of inserting one distinct second, they slightly lengthen many seconds over a window before and after the event. The clock reaches the same corrected time without displaying 23:59:60.
Smearing can reduce software disruption, but there is no single universal smear schedule. Two systems using different methods can disagree by fractions of a second during the window.
Applications requiring precise UTC traceability must document whether they step, smear or use another timescale.
Does GPS use leap seconds?
Satellite-navigation systems maintain their own continuous timescales. GPS time began aligned with UTC in 1980 but does not insert leap seconds in the same way. Receivers use broadcast offset information to convert between GPS time and UTC.
This is one reason navigation and telecom equipment must handle timescales explicitly rather than treating every numeric timestamp as identical.
Are leap seconds going away?
International metrology bodies have agreed on a future change intended to allow UTC and Earth's rotation to diverge more before a correction is required. Implementation details and the eventual method of reconciliation remain matters for international coordination.
For the present, systems should follow current official bulletins and remain capable of handling announced adjustments. A policy planned for the future does not make today's timekeeping assumptions safe.
How software teams prepare
Teams can inventory systems that parse UTC strings, order high-frequency events or depend on external time services. Tests should include the 23:59:60 representation, repeated or smeared timestamps and temporary disagreement between sources.
Critical systems should use monotonic clocks for measuring elapsed duration. A monotonic clock is designed not to jump when civil time is corrected. UTC timestamps remain useful for recording when an event occurred, but they are a poor substitute for a duration timer.
The bottom line
Leap seconds bridge atomic time and the slightly irregular rotation of Earth. They keep civil time astronomically grounded, but the unusual extra second challenges software built on simplified calendars. Explicit timescales, monotonic timers and tested clock-correction policies are the safest response.
For more clear guides, visit the Explainers section and read how news embargo release times work.
Sources: International Bureau of Weights and Measures; International Earth Rotation and Reference Systems Service; International Telecommunication Union.
