World Time Has Been Adjusted Twenty-Seven Times Since 1972. In 2022 the Metrology Conference Voted to Loosen the Rule Behind Them.

by Arjun Mehta · September 2, 2026

Science & Technology Writer, The Globe Ledger

Almost everything that coordinates across borders — satellite navigation, financial timestamping, power grid synchronisation, telecommunications — depends on a single agreed answer to the question of what time it is. That answer is Coordinated Universal Time, and it is not a clock. It is a computed reconciliation between two incompatible things: an ensemble of atomic clocks that keep an almost perfectly uniform second, and a planet whose rotation is neither uniform nor entirely predictable.

The reconciliation is performed by adding a second. The Bureau International des Poids et Mesures, which realises and disseminates UTC, describes UTC as obtained from International Atomic Time by the insertion of leap seconds. Those insertions are what keep civil time tethered to the sky.

Where the rule comes from

The present system starts at a specific instant. A working document of the Consultative Committee for Time and Frequency on the history of UTC records that the defining epoch of 1 January 1972 at 0 hours 0 minutes 0 seconds UTC was set 10 seconds behind International Atomic Time. The same document traces the design further back: the details of the UTC system were worked out by Study Group 7 of the International Radio Consultative Committee in Geneva in 1962 and adopted by the CCIR in its Recommendation 374 of 1963, with the operational rules for leap adjustments approved in 1970 and implemented from 1971.

The procedure itself belongs to a telecommunications recommendation rather than to a physical constant. The Consultative Committee for Time and Frequency, in the document it prepared for the 2022 meeting of the General Conference on Weights and Measures, records that a leap second is applied according to the procedure described in Recommendation ITU-R TF.460-6 of the International Telecommunication Union's Radiocommunication Sector, and that the trigger is a prediction that the difference between UT1, the time scale tied to the Earth's rotation, and UTC is approaching 0.9 seconds.

Deciding when the limit is about to be breached is a separate job again. The International Earth Rotation and Reference Systems Service states in its own glossary that leap seconds are introduced in such a way that UT1 minus UTC stays smaller than 0.9 s in absolute value, that the decision to introduce one in order to meet that condition is the responsibility of the IERS, and that leap seconds are announced in IERS Bulletin C. A CCTF working document on the IERS role, presented to the committee's 20th meeting in September 2015, notes that notification is given roughly six months ahead, well beyond the eight weeks the ITU requires.

Twenty-seven, and all in the same direction

A presentation prepared for the Consultative Committee for Time and Frequency refers to the 27th leap second on 1 January 2017, records the last leap second as 31 December 2016, and gives the difference between International Atomic Time and UTC as 37 seconds. The arithmetic is self-consistent: the 10-second offset fixed in 1972, plus 27 insertions, produces the 37 seconds separating the two scales today.

Counts published at different moments differ, and the discrepancy is one of vintage rather than dispute. An earlier CCTF working document on the IERS role, presented to the committee's 20th meeting at the BIPM on 17 and 18 September 2015, states that since 1972 there have been 26 leap seconds and that all have been positive. The later CCTF presentation counts 27. The two are reconcilable rather than contradictory: the 2015 document was written before the insertion at the end of 31 December 2016, and a reader who cites either figure without its date will be wrong within a year of citing it. Both agree on the more consequential point, which the earlier document states directly: every adjustment made since 1972 has been positive. A second has been added; a second has never been removed.

That asymmetry is now the problem rather than the reassurance. In the accompanying document it prepared for the 2022 resolution, the CCTF raised the possible need for the first negative leap second, an event whose insertion, it noted, has never been foreseen or tested. A negative leap second would require systems to skip from 23:59:58 to 00:00:00, a case that decades of software has had no reason to handle.

Why a second is hard for computers

The CCTF document is precise about the failure mode, and it is not a matter of a second being a long time. Most clocks, and especially digital clocks that keep time as a count of seconds elapsed since some epoch, have no way to represent the time 23:59:60, which is what a positive leap second literally is. Separately, the discontinuity in the measured interval across a leap second is inconsistent with assigning time tags to a real-time process, so any system that assumes elapsed time is monotonic and uniform has a moment each time in which its assumption is false.

Resolution 4 of the 27th General Conference on Weights and Measures, adopted in 2022, states the stakes in institutional language: the introduction of leap seconds creates discontinuities that risk causing serious malfunctions in critical digital infrastructure, including the Global Navigation Satellite Systems, telecommunications, and energy transmission systems.

What was actually decided in 2022

The operative text is short. Resolution 4 decides that the maximum value for the difference between UT1 and UTC will be increased in, or before, 2035. It does not say what the new maximum will be. That is deliberately left open, and the resolution instead requests the International Committee for Weights and Measures to work with the ITU and other affected organisations to propose a new maximum value that would ensure the continuity of UTC for at least a century, to prepare a plan for implementing it by or before 2035, to recommend how often the arrangement should be reviewed thereafter, and to draft a resolution containing those proposals for the 28th General Conference in 2026.

So the decision that has been taken is that the rule will change, and the decision that has not been taken is what the rule will become. Widening the tolerance does not abolish the divergence between atomic time and the Earth's rotation. It relocates the adjustment to a distant future date and, if the century-long continuity objective is met, to a generation of engineers who will inherit a much larger correction to make at once.

UTC is published after the fact

One further feature of the system is worth stating because it is routinely misunderstood. The BIPM describes UTC as obtained from International Atomic Time by the insertion of leap seconds on the advice of the IERS, and states that the values of UTC minus each participating laboratory's local realisation, designated UTC(k), are published at five-day intervals in the monthly BIPM Circular T. Those local realisations are the physical clocks maintained in national metrology institutes and observatories, which send their data to the BIPM. The implication, which follows from the monthly publication schedule rather than from any single sentence in the BIPM's own description, is that UTC is a computed reference assessed after the fact, while the signals a country actually distributes are its own realisation of it.

The practical picture, then, is this: the world's time standard is a monthly reconciliation of a global clock ensemble, tied to the planet by a telecommunications recommendation, adjusted on the advice of an Earth-rotation service, and now set by a metrology conference to be loosened from that planet's rotation in or before 2035. None of those four bodies could deliver the arrangement alone, and none of them can compel any government to adopt it; the CGPM resolves, and the 2022 text does no more than instruct its own committee to consult the ITU and draft a proposal for 2026.