Last week there have been articles published about the most recent (27th) meeting of the CGPM. The most excited headlines were saying that leap seconds were going to stop being a thing starting in 2035 — The NY Times had “It’s Official: The Leap Second Will Be Retired (a Decade from Now)”. More nuanced presentations at Arstechnica had “Network-crashing leap seconds to be abandoned by 2035, for at least a century”.
Since I’ve been writing about timescales, UT1, UTC, leap seconds, and the potential for a negative leap second happening for a while now, I obviously couldn’t leave this alone. So I went to investigate just what the deal was.
What happened
Explaining things gets a bit confusing because there’s a lot of organizations with French acronyms that handle lots of moving parts. So here we go.
The General Conference on Weights and Measures (CGPM) happens every four years and is the supreme ruling authority of the International Bureau of Weights and Measures (BIPM). This organization was formed under the Metre Convention and among the many things they currently do, make the decisions about the SI units. They literally define the units we work with, including the definition of the second that we use to keep track of time with.
Side note: They also adopted Resolution 3, which extended the SI prefixes with ronna, ronto, quetta, quecto, While also adopting Resolution 5, which encourages that a recommendation for a new, improved, definition of the second for the 2026 meeting.
In the context of time, the BIPM manages the International Atomic Time (TAI)— TAI is a weighted average of the time kept by over 450 atomic clocks in over 80 national laboratories worldwide and is the basis for UTC. BIPM essentially takes all the clocks that participate, makes adjustments for altitude, and uses that information to generate the TAI timescale that ticks atomic seconds (i.e. the unperturbed ground-state hyperfine transitions 9,192,631,770 of the caesium 133 atom).
What this means is that TAI ticks according to “atomic seconds”, the ones counted by the cesium clocks around the world according to the SI definition of the second. Incidentally, this second was tuned to roughly equal the “ephemeris second” of the 1820s. That is, if we dragged an atomic clock back to the early 1800s, 1 atomic second would roughly be the same length of duration as 1 mean solar second.
Atomic second?! There’s other seconds?
I’ve previously written a bunch about time scales and leap seconds before already, but here’s the overview. There is, by definition, 86,400 seconds within a day as tracked by atomic clocks and the SI system of measurement. This, effectively, what is being tracked by TAI.
But there is another, alternative, definition of the day — mean solar time. Mean solar time was originally defined using the sun going directly overhead at the Prime Meridian, but is now defined by fixed quasars overhead via astronomical observations because point sources are easier to measure. One day is STILL, again by definition, 86,400 seconds long. This time scale is what’s being tracked by a thing called UT1 (Universal Time 1, there’s also UT0 and UT2 and they’re not important here).
Because the Earth’s rotation has been slowing down over time due to tidal effects and the shifting of mass on and within the Earth’s surface, the actual length of a UT1 second is actually a tiny bit longer than the TAI second. The two definitions of “second” are fundamentally incompatible because they tick at different rates — one constant and one ever slowing.
UTC, is the result of these two time scales being somewhat forcibly reconciled so that the society at large has reliable clock that tracks the sun. UTC is an integer number of seconds offset from TAI and ticks at the exact same rate. UTC is also defined so as to not differ from UT1 by more than 1 second — if it ever drifts more than 0.9s away from UT1 in either direction, a leap second will be added to make sure that the noontime sun will still be overhead at the correct time.
This is why we typically add extra seconds to UTC, UTC’s atomic second ticks faster than UT1 ticks according to the sun.
Okay, I get it! Back to the CGPM Resolution!
At the 27th CGPM, the group adopted Resolution 4, “On the use and future development of UTC”. It’s a fairly short 2-page resolution, but I’ve copied the most relevant bits, and added some extra bold emphasis
recalling that:
…
when the difference (UT1-UTC), as observed by the International Earth Rotation and Reference Systems Service (IERS), is predicted to approach 0.9 seconds, a leap second is applied according to the procedure described in Recommendation ITU-R TF.460-6 of the International Telecommunication Union Radiocommunication Sector (ITU-R),
…
recognizing that the use of UTC as the unique reference time scale for all applications, including advanced digital networks and satellite systems, calls for its clear and unambiguous specification as a continuous time scale, with a well-understood traceability chain,
decides that the maximum value for the difference (UT1-UTC) will be increased in, or before, 2035,
requests that the CIPM consult with the ITU, and other organizations that may be impacted by this decision in order to
propose a new maximum value for the difference (UT1-UTC) that will ensure the continuity of UTC for at least a century,
prepare a plan to implement by, or before, 2035 the proposed new maximum value for the difference (UT1-UTC),
…
encourages the BIPM to work with relevant organizations to identify the need for updates in the different services that disseminate the value of the difference (UT1-UTC) and to ensure the correct understanding and use of the new maximum value.
So to break it down, the CGPM decided that UTC needed to be a continuous time scale, because the discontinuities created by the leap second (and the mere possibility of there being a negative leap second) had caused lots of issues with computer systems in the past and will continue to cause trouble in the future.
But what’s interesting is that it has decided to effectively punt on the problem in a couple of ways.
First, it has decided that the difference of (UT1-UTC) should be allowed to have a new maximum value on or before 2035 and should ensure the continuity of UTC for at least 100 years. People are interpreting this as “we’re not going to insert a leap second whenever UT1-UTC because greater than 0.9s.” They’re going to allow the difference to just keep growing for at least a hundred years (unless they standardize some way to make UTC continuous while somehow keeping UT1-UTC small).
Why wait until 2035? The resolution itself doesn’t say, and I can’t find any meeting minutes or notes. But the article from the NY Times include mentions that it was a compromise due to Russia’s delegation being against an immediate change because their GLONASS satellite navigation system incorporates leap seconds into their clocks, while the other satnav systems like GPS and Galileo never incorporated leap seconds. It probably requires sending up replacement satellite hardware to change the clock systems.
Most importantly, you’ll note that the resolution doesn’t really talk about the dissolution of the leap second! It instead makes room for the leap second to not exist by asking relevant parties to cooperate and figure out a new maximum value for the difference between UT1-UTC (which is currently 0.9 seconds).
So who actually decides what to do with the leap second?
To my surprise, it’s NOT BIPM, nor is it actually the International Earth Rotation and Reference Systems Service (IERS) who is the association of astronomers who do the astronomical observations needed to actually calculate what UT1 is. IERS are the people who effectively say when UT1-UTC has gotten too big and declare that a leap second is needed.
Instead, it is yet another international body, this time within the UN, ITU Radiocommunication Sector (ITU-R), part of the International Telecommunication Union (ITU). In fact, the ITU actually PREDATES the UN and were originally formed via international treaties to coordinate telegraph and radio communication around the world by establishing standards for international communications.
I’m sure you’re wondering what the heck does a radio and telecommunication agency have to do with UTC. Well, the ITU-R, which manages radio communication, established ITU-R TF.460-6 "Standard-frequency and time-signal emissions" because since the 1980s, if not earlier, radio was a primary way for transmitting time signals across countries. ITU-R’s standard has to deal with the transmission of time and a lot of things rely on it.
If you read the 5 page standard, which is extremely sparse on implementation details, it states that UTC is “maintained by BIPM”, with help from IERS to determine when a leap second needs to be added. But then it goes on to describe the allowable tolerances for values of UT1-UTC in transmission, as well as recommendations for how to transmit the time values.
The ITU-R is supposed to meet in 2023 to have a vote on what to do. They had previously decided in their 2015 meeting that the ITU would not be making any decisions about leap seconds until the 2023 meeting.
I don’t think we in the public can see a much of the discussions involving UTC and the handling of the leap second from the ITU’s perspective because messing with UTC apparently starts messing with how countries define important concepts like “day”.
Despite this, the NY Times did have this quote: “Felicitas Arias, the former director of the time department at the B.I.P.M. and now a visiting astronomer at the Paris Observatory, said that negotiations between the two organizations convinced her that the I.T.U. will support the Versailles vote.”
There’s been many years worth of research, discussions, and negotiations about the topic since it was not the first time it’s come up. So there does seem to be hope in the air that we won’t be seeing another leap second after 2035 (if not earlier).
That means that if the Earth is going to spin faster and try to give us a negative leap second, it’s got 12 years to try to do so. (Unlikely.)
What’s this all mean?
Assuming the ITU-R meeting adopts the recommendation of the CGPM late next year, UT1 and UTC will effectively be unmoored from each other for a century in the near-ish future. That means that UT1 will continue to tick ever slower, while UTC marches to its atomic heartbeat. We’ve had 37 leap seconds in the 50 years between 1972-2022, so by 2135, UTC might be a whole minute ahead of UT1 if things stay separated.
The hope is that in the meanwhile, all the relevant agencies, BIPM, ITU, IERS, and world governments, can come to some sort of procedure that will allow UTC to reunite and be in sync with UT1, while remaining continuous. Currently this isn’t possible because it would require the replacement of a TON of time related hardware to adopt some new undefined protocol.
That said, there’s supposed to be a reunification of UT1 and UTC at the end. I suppose in the absolute worst case where no one changes any technology for 110 years, we have a massive >1min leap moment that forcibly yanks clocks back into sync. Sadly I won’t be alive to see happens.
Side note, if you’re asking why can’t we use a Leap Smear like some tech companies are doing. Well, it’s not a standard agreed upon by international convention yet. Plus, it needs to be done to EVERY CLOCK. In the very early days of UTC, leap seconds weren’t a thing and instead time frequencies were often tweaked in partial-second increments on a more frequent basis. People of the time quickly learned that it was an operational nightmare to keep things in sync and the current system was developed.
Maybe reunification of UT1 and UTC happens sooner than 2135 because all the tech and standards are in place before then. We’ll have to see. But until all this stuff is figured out, we as laypeople and computer system users get to just enjoy the the continuous monotonic UTC time scale without worrying about everything else.
Which is pretty darned cool!
Reference Rabbit Hole of Doom
Every time I write about UTC and time, I wind up finding myself at Steve Allen’s site.
Here, specifically, I spent a ton of time reading and sifting through the many broken links at “UTC might be redefined without leap seconds” and a list of events involving the future of leap seconds.
The reason for this is because these international organizations have over a century of history. It’s a GIANT mess of publications, conferences, resolutions, and specifications that evolved over a significant amount of time. It’s more than I can ever hope to unravel for a weekly newsletter post. So I leaned heavily on his work to just barely make sense of the most recent developments.
If you’re looking to (re)connect with Data Twitter
Please reference these crowdsourced spreadsheets and feel free to contribute to them.
A list of data hangouts - Mostly Slack and Discord servers where data folk hang out
A crowdsourced list of Mastodon accounts of Data Twitter folk - it’s a big list of accounts that people have contributed to of data folk who are now on Mastodon that you can import and auto-follow to reboot your timeline
Standing offer: If you created something and would like me to review or share it w/ the data community — my mailbox and Twitter DMs are open.
New thing: I’m also considering occasionally hosting guests posts written by other people. If you’re interested in writing something a data-related post to either show off work, share an experience, or need help coming up with a topic, please contact me.
About this newsletter
I’m Randy Au, Quantitative UX researcher, former data analyst, and general-purpose data and tech nerd. Counting Stuff is a weekly newsletter about the less-than-sexy aspects of data science, UX research and tech. With some excursions into other fun topics.
All photos/drawings used are taken/created by Randy unless otherwise credited.
randyau.com — Curated archive of evergreen posts.
Approaching Significance Discord —where data folk hang out and can talk a bit about data, and a bit about everything else. Randy moderates the discord.
Support the newsletter:
This newsletter is free and will continue to stay that way every Tuesday, share it with your friends without guilt! But if you like the content and want to send some love, here’s some options:
Share posts with other people
Consider a paid Substack subscription or a small one-time Ko-fi donation
Tweet me with comments and questions
Get merch! If shirts and stickers are more your style — There’s a survivorship bias shirt!