Leap second hiatus

November 15, 2020 12:45 PM

Leap seconds exist because the Earth takes (very roughly) about a millisecond more than 24 * 60 * 60 seconds to rotate each day; when we have accumulated enough extra milliseconds, a leap second is inserted into UTC to keep it in sync with the Earth. At the moment the Earth is rotating faster than in recent decades: these shorter days, with a lower length-of-day, means the milliseconds accumulate more slowly, and we get fewer leap seconds.

By Tony Finch.

posted by smcg (18 comments total)

I wonder is there a correlation between global warming's increased temperature and the "shortening" day.

Quick! Someone do a science!
posted by seanmpuckett at 12:54 PM on November 15, 2020 [1 favorite]

There has never been a negative leap second, and if there is one, everyone who deals with NTP or kernel timekeeping code expects that it will be an appalling shitshow.

If there's an argument for the abolition of leap seconds that I could get behind, it's the prospect of a negative leap second.
posted by the antecedent of that pronoun at 12:54 PM on November 15, 2020 [5 favorites]

A negative leap second would probably have horrible problems well beyond NTP and kernel. Though there are monotonic clocks, a lot of software that will break probably does not think to use them.

Which is a pity, because if we could get a negative leap second to work just once, we could avoid future problems: Change to inserting either a positive or negative leap second each year, jittering around the correct value when no change was needed.
posted by joeyh at 1:08 PM on November 15, 2020

I wonder is there a correlation between global warming's increased temperature and the "shortening" day.

That was the first thing I thought: can global cooling/warming change the mass of the Earth such that rotational velocity is affected? Or some other concept that addresses atmospheric density changing the mass? diameter? of the planet.
posted by rhizome at 1:08 PM on November 15, 2020 [2 favorites]

Luckily there's GPS time which ignores leap seconds and is thus drifting away from UTC.

Of course if you are dealing with GPS satellites, earth rotation is the least of your worries since you are also correcting for the effects of special relativity (Satellite clocks are moving thus slower) and general relativity (clocks are higher in the gravity well thus faster).

Of course if you are drifting from UT1 then your satellites are also not where you expect them to be since your sidereal day length is changing so you have to apply these DUT1 corrections anyways.
posted by vacapinta at 1:28 PM on November 15, 2020 [6 favorites]

I don't know why they call it a leap second; I've never leapt during one.
posted by Greg_Ace at 1:43 PM on November 15, 2020 [2 favorites]

Some people just don’t live right I guess.
posted by mhoye at 1:47 PM on November 15, 2020 [3 favorites]

And we thought that sense of time-has-no-meaning-in-2020 was purely subjective.
posted by curious nu at 2:00 PM on November 15, 2020 [5 favorites]

I can't see how kernel or NTP would have much of a problem. Your system's hardware clock already has adjtime(3):

The adjtime() function gradually adjusts the system clock (as returned by gettimeofday(2)). The amount of time by which the clock is to be adjusted is specified in the structure pointed to by delta.
...
If the adjustment in delta is positive, then the system clock is speeded up by some small percentage (i.e., by adding a small amount of time to the clock value in each second) until the adjustment has been completed. If the adjustment in delta is negative, then the clock is slowed down in a similar fashion.


NTP works the same way, as long as your clock is withing a few seconds of the NTP time (calculated from 3 or more sources) the system clock time is slowly adjusted over a period of time to match. It's only if the system clock is far away from NTP time that you have to do a force-set before the automagic clock training sets in.
posted by zengargoyle at 2:03 PM on November 15, 2020 [2 favorites]

I was just explaining to the kids that what we think of as “the time” is just seconds since January 1 1970. 00:00:00 UTC and that we’ve only had Coordinated Universal Time since Jan 1, 1960. As we drift back before railroads in the 1800s we lose all coordination of time beyond some local estimate of the solar noon for a town village. Eventually we can go back to July 6, 1169 and the reign of Richard the First of England a date selected in 1275 by the Statute of Westminster as end of “time immemorial” and the beginning of modern time record keeping as far as the British were concerned and since they had that big empire thing the whole world ended up cursed to forever live under their notions of being on time.
posted by interogative mood at 2:35 PM on November 15, 2020 [10 favorites]

Not we keep launching rockets into space, maybe. Maybe that's another thing Elon Musk can put on his resume: "Reusable rockets contributed to the conservation of Earth mass to ensure the consistency of UTC."
posted by ardgedee at 4:45 PM on November 15, 2020

Don't forget the time the British calendar skipped 11 days in 1752.
posted by fings at 5:02 PM on November 15, 2020 [1 favorite]

Probably more linked to conservation of angular momentum and moving water downhill into the ocean.
https://www.livescience.com/53071-melting-glaciers-change-earth-spin.html
https://advances.sciencemag.org/content/1/11/e1500679
posted by Nitro Status at 6:22 PM on November 15, 2020

Negative leap seconds don't actually involve the clock going non-monotonic -- they make the clock go from 23:59:58.999 to 00:00:00 (skipping the second of 59).

I'm sure they would break plenty of stuff that didn't have test coverage for them, but at least they don't introduce never-before-seen state onto your clock. Seeing a digital clock reading :60 is weird. Breaks a lot of people's time formatting and parsing code too.

If you implement leap second smearing (ntpd, Google NTP), then clients of time see no fundamental difference between positive and negative leaps.
posted by away for regrooving at 9:07 PM on November 15, 2020

The earth looses about 55,000 tonnes of mass annually. The approximately 65 tonnes of mass lost annually since the mid 20th because of escape velocity space launches (which are the only ones that count for this purpose) account for less than 4000 tonnes in total, a barely measurable fraction.
posted by Mitheral at 6:22 AM on November 16, 2020 [3 favorites]

The earth looses about 55,000 tonnes of mass annually.

"Loose the space mass!"

"But...master...won't that risk a negative leap second?"
posted by straight at 10:18 AM on November 16, 2020 [2 favorites]

Mitheral, does that figure include the approxately 100 metric tons of space dust that impacts the Earth daily?

(Herzog's new _Fireball_ has spectacular closeups of it.)
posted by joeyh at 11:58 AM on November 16, 2020

It's the net. The numbers are from wiki which says: The 5.5×107 kg annual net loss is essentially due to 100,000 tons lost due to atmospheric escape, and an average of 45,000 tons gained from in-falling dust and meteorites.

Mass loss is due to atmospheric escape of gases. About 95,000 tons of hydrogen per year (3 kg/s) and 1,600 tons of helium per year are lost through atmospheric escape. The main factor in mass gain is in-falling material, cosmic dust, meteors, etc. are the most significant contributors to Earth's increase in mass. The sum of material is estimated to be 37000 to 78000 tons annually, although this can vary significantly; to take an extreme example, the Chicxulub impactor, with a midpoint mass estimate of 2.3×1017 kg,[36] added 900 million times that annual dustfall amount to the Earth's mass in a single event.
posted by Mitheral at 12:46 PM on November 16, 2020 [1 favorite]