Leap seconds
-
Day 26: The Astronomical Anchor Problem
If Einstein’s gift to us yesterday was that there is no universal “now,” today I want to come back down to Earth and revisit something more practical: how we measure time, and how we define the second.
We dug into the second itself back on Day 8. The short version is that there are two competing definitions, one astronomical and one atomic, and since 1972 they’ve drifted about 37 seconds apart. The gap is small, but it grows, and the system we built to manage it (leap seconds) is being dismantled. Today I want to look at what that means, and whether we should replace it with anything.
As you read this, I want you to sit with one question: what is civil time actually for? This isn’t a technical problem, it’s a values problem. What we anchor civil time to depends entirely on what we decide it’s for, and our civilization has only recently started changing its mind about that, mostly without noticing. More people should be paying attention.
Two definitions of the second
There are two ways to define a second.
The first is astronomical. A day is one rotation of the Earth, a second is 1/86,400 of that, and you divide the sun-up-sun-down cycle into uniform pieces. This is UT1, the modern name for “time measured by Earth’s rotation.” It’s what humans used for most of history.
The second is atomic. Since 1967 the official second has been 9,192,631,770 transitions between two energy levels of cesium-133 (we covered this on Day 8). This is TAI, International Atomic Time, and it’s defined without any reference to the Earth at all.
We know the Earth is a bad clock. Its rotation slows over geologic time, wobbles with the atmosphere and oceans, and drifts unpredictably on decade scales for reasons involving the molten core that we don’t fully understand. The astronomical second is not the same length from one day to the next.
The atomic second is, by construction, exactly the same length. It’s the most precisely defined quantity in human civilization, and it does not care what the Earth is doing.
So the atomic second became the official one, and UT1 has been drifting against it ever since. We invented UTC, the civil time on your phone, as a compromise: tick at the atomic rate, but insert a leap second whenever the gap from UT1 approaches 0.9 seconds. I wrote about the mess that caused on Day 19 and Day 20.
In 2035, leap seconds are being abolished… probably. After that, UTC will be allowed to drift from solar time by somewhere between one minute and one hour, to be decided by a 2026 vote. The astronomical anchor is being loosened, and maybe eventually cut entirely.
What “anchored to the sun” actually means
Civil time should be anchored to Earth’s rotation means two completely different things depending on who says it.
If a chronobiologist says it, they mean morning light should arrive at roughly the clock time when people wake up. That’s a circadian-health argument, the mechanism is real, and I dug into it on Day 21. But it’s an easy bar to clear: civil time only has to stay within about an hour of the sun. We already break that constantly. Time zones and daylight saving routinely shove the clock an hour or more off solar noon and mess with everyone’s circadian rhythm far more than a slow drift ever would, and society carries on. If anything, that’s a better argument that DST is dumb than that the anchor matters.
If an astronomer says it, they mean something more literal. Picture the Earth as a spinning sphere: at any instant you can calculate its orientation relative to the sun, and for most of history that was time. The official clock pointed at the Earth’s angular position because that’s all “time” ever meant. It’s a clean, logical definition, and it made complete sense for thousands of years. But it’s a definitional and cultural argument, not a biological one, and where the chronobiologist’s bar is easy to clear, this one is strict and expensive. With atomic clocks keeping time and UT1 tracking the Earth separately, anchoring civil time to that geometry makes far less sense now than it did at the dawn of civilization.
The metrologists' argument is that we can drop the astronomical anchor without losing anything humans actually depend on. We’ve all grown used to solar noon being noon, and that attachment is understandable, but in their view it’s swappable: let it go and you still have a stable, well-defined, engineering-grade measurement of time.
I don’t think they’re crazy. The tradeoff is real: we accept that civil time slowly comes unmoored from its position in the sky, and in exchange we get a clock that never needs correcting.
What is not decided yet
There are decisions still on the table. The 2022 vote killed the 0.9-second tolerance by 2035 but left the replacement open. After three years of the standards bodies working through options, the call lands this October: the 28th meeting of the CGPM, the General Conference on Weights and Measures, convenes in Versailles, France, and one item on the table is the new limit, how far UTC will be allowed to drift from solar time. The candidates range from keeping it nearly as tight as today to abandoning the anchor completely:
- 1 minute (no correction needed for about a century)
- 1 hour (a correction roughly every several thousand years)
- No limit at all (let civil time drift from the sun forever)
There is no neutral position here. The choice is really a vote on how much we still care about the sun.
Either way, astronomers lose the easy version of this. Today UTC doesn’t stray much from Earth’s rotation, close enough to read orientation straight off the civil clock. Once the tolerance loosens to a minute or more, that stops being true under every option on the table, so we will have to wire up UT1 directly to its own time signal.
The clock keeps clocking
Future generations will probably look back and ask why we did it this way. The honest answer is that time is complicated. Leap seconds made sense for part of our history, and they probably don’t make sense forever. The clocks will keep clocking either way.
Tomorrow: a survey of the prior attempts at fixing civil time. It has happened before, and we can learn a lot from how they failed.
Sources
- Atomic vs. Astronomical Time (TAI & UT1): International Earth Rotation and Reference Systems Service (IERS) and BIPM definitions.
- The Abolition of the Leap Second: Resolution 4 of the 27th General Conference on Weights and Measures (CGPM), 2022.
- Who is deciding the new tolerance: The CCTF Task Group on a continuous UTC, established by the BIPM in 2023 to draft the new maximum UT1−UTC value for the 28th CGPM (2026).
- Where it gets decided: The 28th General Conference on Weights and Measures, Versailles, 13–15 October 2026 — the meeting set to vote on the new maximum UT1−UTC value.
- The one-minute proposal: Levine, Tavella & Milton, “Towards a consensus on a continuous Coordinated Universal Time,” Metrologia, 2022 — argues a tolerance of about one minute keeps UTC within UT1 for roughly a century.
I’d appreciate a follow. You can subscribe with your email below. The emails go out once a week, or you can find me on Mastodon at @[email protected].