Mark your calendars: celebrate the Leap Second!

Tomorrow, a leap second will be introduced. That means that one minute in the last day of 2005 will be 61 seconds long. This is the first time a leap second has been needed in seven years.

According to Bulletin C of the International Earth Rotation Service:

The sequence of dates of the UTC second markers will be:

2005 Dec 31,  23h 59m 59s
2005 Dec 31,  23h 59m 60s
2006 Jan 01,  00h 00m 00s

The difference between UTC and the International Atomic Time TAI is:

from 1999 Jan 1, 0h UTC, to 2006 Jan 1 0h UTC : UTC-TAI = - 32s
from 2006 Jan 1, 0h UTC, until further notice : UTC-TAI = - 33s

In other words, the year will be longer by one second just before midnight, new year's eve, GMT. That's 4:00 PM PST, 7:00 PM EST.

Please take that second to just go nuts. One second. All to yourself. Use it wisely.

I also recommend reading Paladin of the Lost Hour by Harlan Ellison.

The reason leap seconds are needed is because there are two different definitions of the length of a second: the solar time that you use every day, where the there are 60×60×24 seconds per day (meaning the length of the second varies as the speed of rotation of the earth varies) and the engineering definition, where a second is defined as 9,192,631,770 oscillations of a cesium-133 atom: the atomic clock.

Mostly humans are concerned with time as it relates to the comings and goings of the Sun, but you need to be a little more accurate if you expect that satellite to be where you left it, so you need the length of your unit of measure to not change. That would affect everything else, too, since the official definitions of other units like the volt and the meter are based on the second. (You thought the meter was still defined by the length of that platinum-iridum bar in France, didn't you? Get with it!)

So the time system we use day to day, UTC ("Coordinated Universal Time") is a compromise between solar time (UT1, "Universal Time", formerly GMT, "Greenwich Mean Time") and atomic time (TAI, "International Atomic Time"): in UTC, seconds are of a fixed length, but leap seconds are periodically introduced to prevent "solar" noon and "atomic" noon from drifting away from each other.

But really it's all a lot more complicated than that!

And then of course there was the switch from the Julian to the Gregorian calendar at various dates (mostly) between 1582 and 1752, in which 4-Oct-1582 was immediately followed by 15-Oct-1582, in order to correct for the inaccurate leap-day-insertion algorithm in the Julian system, which over the centuries since its introduction had caused the calendar to detach from the observed equinoxes by ten days.

It can be really hard to know what time it is.

Tags: , ,

39 Responses:

  1. wfaulk says:

    I love why "Coordinated Universal Time" (or, in French, "temps universel coordonnè") is abbreviated UTC: so that it wouldn't be accurate in either French or English.

    • brad says:

      Yeah, I love international compromises. Just like how (iirc?) ATM's frame size was a split between 64 bytes and 32 bytes. Japan wanted 32 bytes to reduce latency, so they wouldn't have to deal with adding echo cancellation hardware, but the US, a big country with lots of distance/latency already had echo cancellation everywhere, so they wanted more bandwidth. So ATM frames are like 53 bytes.

    • smackfu says:

      We should change the English name just to spite them: Universal Time (Coordinated)

      • pne says:

        Which is, apparently, roughly where the term "UTC" comes from.

        At any rate, an explanation I've seen indicates that, like UT1, it's a kind of UT, specifically a coordinated UT -- using a subscript, as in UTC captures the idea. Only they use normal letters: UTC.

        And talking of compromises between English and French: having the International Organisation for Standardisation have the short name ISO is also interesting... though I believe they claim it's not an acronym but is derived from the Greek morpheme iso-.

        • psr says:

          I read somewhere that iso is so named because it means "the same" not only in English and French, but also in Russian.

    • I wanted to punch someone when I heard the explanation of ISO. Like most people, I thought it was the International Standards Organization. Nope. It's the International Organization for Standardization. And ISO isn't even an acronym. It's "iso" as in the Greek derived prefix "iso-", as in isometric. Meaning the same.

  2. fo0bar says:

    Remember, if you live in London (or any place UTC +00:00), you must begin the new years countdown at 23:59:51.

    (You'll be the hit of the party when you try explaining that to the other partygoers.)

    • cdamian says:

      or you count 10 9 8 7 6 5 4 3 2 1 1 "Happy New Year".

      or is it 10 9 8 7 6 5 4 3 2 1 "Happy New Year" "Happy New Year" ? eg twice the amount of kissing random strangers and drinking too much booze.

      here in spain everyone has to eat a grape with each of the twelve sounds of the clock, which can be very tricky.

    • fgmr says:

      You'll be the hit of the party when you try explaining that to the other partygoers.
      Heh, just like you were the hit in 1999 when you tried to explain that the big celebration should be a year later.

  3. ritcey says:

    "It can be really hard to know what time it is."

    Everytime I attempt to write a program that deals with time/dates, I'm reminded of this fact. It just ends up feeling so... arbitrary.

  4. lars_larsen says:

    This is why I dont wear a watch.

  5. edge_walker says:

    (meaning the length of the second varies as the speed of rotation of the earth varies)

    For a great explanation why the speed varies (and how the tides really work, which is often incorrectly explained), check Moon Mechanics: What Really Makes Our World Go 'Round.

  6. gnat23 says:

    The real question: what will you do with your extra second?

  7. davel_jonez says:

    Whatever I'm doing tomorrow, I'm going to use that second to put my hand down my pants.

  8. prog says:

    I am delighted to learn that there is an organization known as the International Earth Rotation Service. It sounds like, besides publishing calendar recalibration advisories and such, they also take semiannual trips to the poles to solemnly re-oil the axis and make sure everything's bolted down nicely.

    • jwz says:

      It is, truly, one of the best names ever.

      • jwjr says:

        The whole system is full of wonderful names. The US part, which participates through the US Naval Observatory, is the "US Earth Orientation Service". I think any "Earth Orientation Service" is a close second to an "Earth Rotation Service".

        Also splendid, in a different sense, is the name of the division that actually figures out what's happening right now for people keeping track of satellites or the like: "IERS Rapid Service/Prediction Center for Earth Orientation Parameters". It sounds like it came straight out of 1940s science fiction.

        "Where is our satellite at this very moment, Chet?"

        "Hang on one minute boss-- once I get the boys at the Rapid Service Prediction Center on the horn to find out the latest Earth orientation parameters, I should be able to tell you down to the inch!"

  9. mysterc says:

    How many IRS agents visit themevery year when they put their occupation down on their taxes?
    "No, really... I am time cop, in a manner of speaking."

  10. The definitive, doorstop-sized reference for all this is the modestly titled 'Explanatory Supplement to the Astronomical Almanac'. Eight hundred pages of the actual details.

  11. spike says:

        I wonder how many programmers didn't know that struct tm.tm_sec can == 60.
        Sadly, we may find out.
        Or perhaps more sadly, we won't.

    Either way, enjoy the extra time tonight!

    • cabrius says:

      Earlier this year I did some code that did some sanity checking on various bits of data, and for a date/time type I let the seconds go up to 61, as permitted by localtime(3) (on Linux, anyway).

      It took me a while before I noticed that a coworker had changed it to a plain 0-59 check, so I had to change it again and added a comment this time.

      I guess it wasn't as common knowledge as I thought...

  12. chetfarmer says:

    I didn't know the title or author at the time, but I saw "Paladin" on TV once. It also exists as a first-season episode of the mid-80s "Twilight Zone" revival. Danny Kaye is the old man.

  13. As I was walking down the street one day, a man came up to me and asked me what the time was that was on my watch, yeah, and I said; "Does anybody really know what time it is?"
    You made me quote Chicago, god damn you to hell.

  14. airmax says:

    ever exist in all IT - date/time conversions.

    For the f. reasons shouldn't we just divide simple "UNIX ticks" by 60 twice to get the real time? huh?

    If time value floats, then also the volt and amper do, and are those pedantic dummies REALLY sure that incrementing the machine clock once a seven years get them REAL value, not distorted by other physical values and constants floating as well.

    stupid and error prone strategy

  15. jp_larocque says:

    I am happy to report that the service correctly displayed 23:59:60.

    syslogd—POSIX really—fails it:

    Dec 31 15:59:59 evanescence kernel: Clock: inserting leap second 23:59:60 UTC

  16. spendocrat says:

    At my Local University the first-year phyics lab manual - in an attempt to enlighten those few students who actually read it - opens the chapter on accuracy and precision with the text

    A man with one watch always knows the time. A man with two watches is never sure.

  17. It can also be really hard to know what a Planet is. I discovered this over Christmas break, whilst idly wondering whether that whole "is Pluto still a planet" thing had been figured out. It's way more complicated than I expected.