The variable rate of the Earth’s rotation

Recall how in one of my recent posts on the radiation paradoxes, I spoke about how you can measure the spinning rate of the Earth by looking at the stars and also by measurements taken purely on the Earth and that these two methods produce results that are remarkably close. This was used in support of the claim by Bishop Berkeley that it was the stars that exerted a dynamical influence on the Earth and that our motion was relative to, not space itself as Newton thought, which he felt was an unobservable entity and thus had no relevance.

The rate at which the Earth is spinning on its axis is not fixed though. Over time it has been slowing down, meaning that the days have been getting longer. Around 600 million years ago, the day was about 21 of our present hours. But a new wrinkle appeared in the last half century in that the rate of rotation was increasing slightly and this caused problems. As our ability to measure time became more accurate with the adoption of atomic clocks, this required the regular adoption of the so-called ‘leap second’, which was a second added to clocks to bring them back into sync with the time as measured with respect to the stars.

Earth is rotating faster than it has in the last half-century, resulting in our days being ever-so-slightly shorter than we’re used to. And while it’s an infinitesimally small difference, it’s become a big headache for physicists, computer programmers and even stockbrokers.

Hundreds of millions of years ago, Earth made about 420 rotations in the time it took to orbit the Sun; we can see evidence of how each year was jam-packed with extra days by examining the growth lines on fossil corals. Although days have gradually grown longer over time (in part because of how the moon pulls at Earth’s oceans, which slows us down a bit), during humanity’s watch, we’ve been holding steady at about 24 hours for a full rotation — which translates to about 365 rotations per trip ’round the Sun.

In the 1950s, scientists developed atomic clocks that kept time based on how electrons in cesium atoms fall from a high-energy, excited state back to their normal ones. Since atomic clocks’ periods are generated by this unchanging atomic behavior, they don’t get thrown off by external changes like temperature shifts the way that traditional clocks can.

Over the years, though, scientists spotted a problem: The unimpeachably steady atomic clocks were shifting slightly from the time that the rest of the world kept.

“As time goes on, there is a gradual divergence between the time of atomic clocks and the time measured by astronomy, that is, by the position of Earth or the moon and stars,” says Judah Levine, a physicist in the time and frequency division of the National Institute of Standards and Technology. Basically, a year as recorded by atomic clocks was a bit faster than that same year calculated from Earth’s movement. “In order to keep that divergence from getting too big, in 1972, the decision was made to periodically add leap seconds to atomic clocks,” Levine says.

The new problem is that the rate of spinning is now slowing down again which might require the subtraction of a second, and that may cause major headaches.

But a negative leap second would present scientists with a whole new set of challenges. “There’s never been a negative leap second before and the concern is that software that would have to handle that has never been tested operationally before,” Whibberley adds.

Whether a regular leap second or a negative leap second is called for, in fact, these tiny changes can be a massive headache for industries ranging from telecommunications to navigation systems. That’s because leap seconds meddle with time in a way that computers aren’t prepared to handle.

“The primary backbone of the internet is that time is continuous,” Levine says. When there’s not a steady, continuous feed of information, things fall apart. Repeating a second or skipping over it trips up the whole system and can cause gaps in what’s supposed to be a steady stream of data. Leap seconds also present a challenge for the financial industry, where each transaction must have its own unique time stamp — a potential problem when that 23:59:59 second repeats itself.

There are those that argue that we should abandon this tinkering and just let the two systems drift further and further out of sync until a major time shift is needed.

Levine says he thinks that leap seconds might not be worth the trouble they cause: “My private opinion is that the cure is worse than the disease.” If we stopped adjusting our clocks to account for leap seconds, it could take a century to get even a minute off from the “true” time recorded by atomic clocks.

Still, he concedes that while it’s true that time is just a construct, a decidedly human attempt to make sense of our experiences in a big, weird universe, “it’s also true that you have the idea that at 12 o’clock noon, the Sun is overhead. And so you, although you don’t think about it often, do have a link to astronomical time.” Leap seconds are just a tiny, nearly invisible way of keeping that link alive.

It was only within the last two centuries, with the advent of more rapid travel between places, that it became important to synchronize time across the globe and create time zones and the like. Until then, time was kept locally, with each place setting it according to the noon day sun. It is not clear how different atomic clock time and astronomical time could get before it becomes something that has a significant effect on our lives.


  1. Bruce says

    This problem has already been solved, by legislatures.
    They often hold a vote for say 20 minutes, and then refuse to note the passage of time until they have the votes in. Or, the legislative session legally ends at midnight, but nobody is allowed to look at the official clock until 3 am.
    There is nothing wrong from a computer point of view if 23:59:59 lasts for two or three seconds instead of one second. With appropriate rounding, how could they tell the difference between 23:59:59.01 and 23:59:59.321?
    It is analogous to when the version of Mac OS X went from 10.9 to 10.10 to 10.11 etc. we all thought the number after the decimal was a single digit that had to turn over once 9 is over, until that was no longer true. Just tell all the clocks that we are shifting to hexadecimal time for a few seconds.

  2. Ken Baker says

    “The new problem is that the rate of spinning is now slowing down again which might require the subtraction of a second, and that may cause major headaches.”

    Wait a sec (see what I did there?) wouldn’t a slowing rate necessitate adding a second?

  3. Peter B says

    Make the adjustment a microsecond at a time every minute until done. Unless I bumbled it will be done in somewhat less than two years.

  4. Pierce R. Butler says

    From the second link in the OP:

    Since the first leap second was added in 1972, scientists have added leap seconds every few years. They’re added irregularly because Earth’s rotation is erratic, with intermittent periods of speeding up and slowing down that interrupt the planet’s millions-of-years-long gradual slowdown.

    “The rotation rate of Earth is a complicated business. It has to do with exchange of angular momentum between Earth and the atmosphere and the effects of the ocean and the effect of the moon,” Levine says. “You’re not able to predict what’s going to happen very far in the future.”

    But in the past decade or so, Earth’s rotational slowdown has … well, slowed down. There hasn’t been a leap second added since 2016, and our planet is currently spinning faster than it has in half a century. Scientists aren’t sure why.

    If atmospheric and oceanic effects make a difference, we should probably look first to global warming (in the absence of external gravitational bodies, etc). I understand why scientists don’t want to take chances on making possibly erroneous conclusions, but what else could make this difference?

  5. birgerjohansson says

    The reindeer above the Arctic circle have long since solved this problem.
    During the arctic night when the sun never rises they simply decouple their activities from the 24-hour cycle.

  6. EigenSprocketUK says

    Leap seconds already run up to 23:59:60 and then roll over to the next day. Negative leap seconds would roll over after 23:59:58 instead.
    How any given server or client responds is the tricky bit.
    I heard that (¿Google?) services work on a fudged network time which runs slow for a period of time. Services which are already regularly resyncing themselves would just not notice. Same would apply to running fast for a negative leap second.
    But microsecond-accurate logs will not be so forgiving (eg financial trades).
    I think there’s an interesting Tom Scott video about a major fraud relating to cheating the time signal.

Leave a Reply

Your email address will not be published. Required fields are marked *