What is a leap second, and who decides to insert one
Key facts
- A leap second is one extra second inserted into UTC so that the scale stays close to the rotation of the Earth. [1]Official
- 27 leap seconds have been inserted since 1972, and the most recent one took effect on 1 January 2017. [2]Official
- TAI minus UTC has been 37 seconds since that date, so UTC minus TAI is −37 seconds. [2]Official
- The IERS Earth Orientation Centre publishes the file of every value of TAI minus UTC, updated through IERS Bulletin 72 of July 2026. [2]Official
- The 27th CGPM decided in 2022 that the maximum value for the difference UT1 minus UTC will be increased in, or before, 2035. [3]Official
A leap second is one extra second added to Coordinated Universal Time. It is added so that clock time does not drift away from the turning of the Earth [1]Official.
It is a repair, not a rhythm. Nothing in the calendar asks for it, and nothing in an atomic clock produces it. A human decision puts it there.
Why an extra second is needed at all
Two different things are called time here. One is counted by atomic clocks; the other is the angle of a spinning planet.
International Atomic Time, TAI, is the weighted average of hundreds of atomic clocks. UTC is derived from it and differs from it by a whole number of leap seconds [1]Official.
The Earth does not rotate at a fixed rate. Without a correction, a scale built from atomic clocks would slowly move away from the position of the Sun in the sky, so whole seconds are inserted to hold the two together [1]Official.
The scales involved, and who keeps them
Four scales appear in every leap second discussion. They are not rivals: each one measures something different, and each one has an owner.
| Scale | What it measures | Who keeps it | Source |
|---|---|---|---|
| TAI | elapsed time counted by a weighted average of hundreds of atomic clocks | the BIPM, from clock data sent by national laboratories | [1]Official |
| UT1 | the rotation angle of the Earth itself | the IERS Earth Orientation Centre at the Paris Observatory | [2]Official |
| UTC | atomic time kept within a stated limit of the Earth’s rotation, by whole leap seconds | the BIPM, with the leap second decision taken at the IERS | [2]Official |
| TT | the theoretical scale that TAI realizes, offset from TAI by 32.184 s exactly | defined by the IAU, realized through TAI | [4] |
The difference UT1 − UTC is the quantity a leap second controls. When the gap grows too large, a second is inserted and the gap drops back by one second [3]Official.
Who announces a leap second
The decision belongs to the International Earth Rotation and Reference Systems Service. The IERS announces each leap second, or the absence of one, in its Bulletin C, and it maintains the machine-readable file that carries every value of TAI − UTC since 1972 [2]Official.
That file is the primary record. Its current edition is updated through IERS Bulletin 72, issued in July 2026, and it expires on 28 June 2027 [2]Official.
Nothing else announces leap seconds. Operating systems, phones and time servers copy the IERS decision; they do not make one [4].
Why leap seconds do not follow the calendar
A leap second is not the calendar’s leap day. A leap day follows a rule you can compute for any year. A leap second follows a measurement of the planet.
The intervals are irregular. Seconds were inserted almost every year in the 1970s, then once in 1999 and not again until 2006 [2]Official. The full sequence is set out in every leap second since 1972.
Two insertion points exist by convention: the end of June and the end of December. Which of them is used, and whether either is used at all, is decided each time [2]Official.
What is decided, and what is only proposed
This is where reporting usually goes wrong, so the two statements are worth keeping apart.
Decided: the 27th CGPM, in Resolution 4 of 2022, decides that the maximum value for the difference UT1 − UTC will be increased in, or before, 2035 [3]Official. A larger limit means leap seconds stop being needed as often.
Proposed: Decision CIPM/115-5 of 2026 endorses an update of Draft Resolution C for the 28th meeting of the CGPM, fixing 20 May 2027 as the implementation date of the new maximum value [5]Official. That is a proposal on the agenda of a meeting held on 13–15 October 2026 [6]Official. No vote has been taken.
The size of the new limit is not stated in Resolution 4 and not stated in the published CIPM decisions [3]Official. Press coverage describes the change as a move towards a leap hour [7], but that figure is press wording, not a published decision.
Whether the next one has a date
It does not. The question is answered in full, with the current offset and the state of the October agenda, in when is the next leap second.
Why a site about lunar time writes about this
UTC has two jobs: count seconds, and stay close to the Earth’s rotation. The second job is the one that produces leap seconds.
A lunar scale has no equivalent job. Nothing in the published lunar proposals ties a lunar scale to the Moon’s rotation, which is the step without a lunar counterpart in how UTC is calculated. That is why our own model of lunar time carries no leap seconds at all and says so on the methodology page, where every model value we print is derived [4].
Where UTC sits among the other scales is drawn in the family tree of time scales, and the Learn hub lists everything else we have written.
Sources
- Policy on Celestial Time Standardization
- Value of TAI-UTC (IERS leap second data file)
- Resolution 4 of the 27th CGPM (2022): On the use and future development of UTC
- Our calculation: the lunartime.org clock model
- Meetings of the CIPM: Outcomes
- Convocation of the 28th meeting of the CGPM (O/Ref AK-CFA-2026-1001)
- Exclusive: International timekeepers to vote on changing the leap second to a leap hour
Last verified