acanuck's picture

    Weighing whether to wait a second, scientists put off decision for three years

    I’m sure you are all as relieved as I am that Thursday’s meeting of the International Telecommunications Union postponed its scheduled vote on whether to drop the leap second:

    The next planned one-second adjustment to Universal Time, at the end of June, will go ahead. And delegates will return home for consultations before the issue arises again at the World Radio Conference in 2015.

    The Americans, French and Japanese are reportedly leading the charge for abolition, while China, the U.K. and Canada are among those opposed. Me too, although I don’t get to vote.

    We need to keep the leap second because otherwise we begin to lose contact with, and understanding of, what we set out to measure in the first place: time as a measure of human experience. Not to get too hyperbolic, the cycles of day and night, spring, summer, fall and winter, life and death.

    Up till the 1950s, Greenwich Mean Time set the world standard for precisely measuring time, and it did so through direct observation of Earth’s rotation and motion relative to the sun. Then atomic clocks were developed that could keep time very consistently around the world, independent of actual solar observations. In 1967, the length of the second was defined as a precise number of “transitions” (no, I have no clue either) of a caesium atom.

    The problem that emerged (and that people knew existed before adopting the atomic standard) is this: the yardstick was more precise than what we were trying to measure. Earth’s rotation varies very slightly year to year, due to things like tides, glacial rebound and earthquakes, and in any case has been slowing down since the planet’s formation. The leap second was introduced in 1972 as a way to keep Universal Time (atomic time) synced with GMT (till then, purely solar time). An extra second would be inserted (or dropped) in any year calculated as running long (or short). The resulting adjustable measure was called Co-ordinated Universal Time (UTC), for all purposes now interchangeable with GMT.

    One argument cited at this week’s ITU meeting for wanting to drop leap seconds is that they occur whenever the difference between UTC and mean solar time nears a full second -- in other words, at irregular, unpredictable intervals. (Eleven seconds were added in the period from 1972 to 1983, but none in the period 199-2004. Messy.)

    That adds unwanted complexity and (theoretical) possibility of error to some navigation or communication systems, like satellites in orbit or interplanetary probes. GPS and its Russian counterpart, GLONASS, for example, ignore leap seconds in their internal operations. (But having already opted out, the global positioning providers aren’t the ones pushing for a change.)

    Agreed, skipping leap seconds would have virtually no affect on anyone. At the current rate of slippage, it would take about a century to accumulate a minute’s divergence from observed solar time, roughly 6,000 years to total one hour. Proponents of the change have seriously suggested that we simply wait for that to happen, then insert a “leap hour.” In 6,000 years or so!

    I have my own suggestion for the time-masters: you can drop the leap second if you first redefine the length of the second. Otherwise, stick with the status quo.

    As of June, there will have been a total of 25 seconds inserted in the 40 years we’ve followed the current atomic definition; not one second has been removed. That’s an upward adjustment every two out of three years. Pretty clearly, the baseline used to initially calculate the length of the second was decades out of date; most of the correcting that’s now being done is for slowing that had already occurred by the 1960s.

    Build that two-thirds of a second per year (or half that much if you’re chicken) into a recalculated atomic definition. Don’t even worry if it ends up proving high. Earth’s rotation is inexorably slowing down; eventually it will fall into sync. Just wait a century or two or three and you’ll look like geniuses.

    My bottom line is this: yes, in this e-linked world we need consistency and accuracy in timekeeping. And yes, defining the second as 9,192,631,770 cycles of a caesium 133 transition does give consistent, repeatable results. But unlike, say, distance – which we can and do assign arbitrary measuring sticks to – time on Earth comes with its own built-in measures: the day, the year.

    And by built-in I mean built into us humans. The week, the hour, the minute and the second may seem arbitrary, but they are just names for divisions or multiples of those natural measures.

    We’ve decided to call one-86,400th of a day a second – and that real definition has to take precedence over a half-century-old (and apparently low-balled) estimate of how many times a caesium atom can undergo a particular change in that amount of time. Sorry, but the cycles of Sun and Earth, even if they are fuzzy and changeable, must define time on this planet. Thinking that its measurement can be decoupled from them is simply perverse.

    (Apologies if this post ran a bit long. I arbitrarily inserted a couple of leap paragraphs.)


    Let me start the backtracking immediately. No, I'm not a scientist. Yes, I've grossly oversimplified the issue. Off the top of my head, I could list about two dozen different measures of time, other than UTC and mean solar, that aren't even mentioned in this post. All irrelevant to the point I'm making.

    I like it. It makes a lot of sense. I'm curious—did you come up with this idea yourself? It's not one I've heard before.

    As for distance, in case you didn't know (I suspect you do), the meter is now defined as the distance light travels in 1/299,792,458 seconds. You might call that arbitrary; I just call it poetry. Since a second is (currently) defined as "the duration of 9,192,631,770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the caesium 133 atom", that allows for an alternate definition: how far light travels in 656,616,555/21,413,747 (approximately 30.663319) "periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the caesium 133 atom". That latter definition, of course, would make it safe for redefining the second.

    Like the word "hyperfine."  Had to look it up (does it answer my question below?)

    The term hyperfine structure refers to a collection of different effects leading to small shifts and splittings in the energy levels of atoms, molecules and ions. The name is a reference to the fine structure which results from the interaction between the magnetic moments associated with electron spin and the electrons' orbital angular momentum. Hyperfine structure, with energy shifts typically orders of magnitude smaller than the fine structure, results from the interactions of the nucleus (or nuclei, in molecules) with internally generated electric and magnetic fields.

    In atoms, hyperfine structure occurs due to the energy of the nuclear magnetic dipole moment in the magnetic field generated by the electrons, and the energy of the nuclear electric quadrupole moment in the electric field gradient due to the distribution of charge within the atom. Molecular hyperfine structure is generally dominated by these two effects, but also includes the energy associated with the interaction between the magnetic moments associated with different magnetic nuclei in a molecule, as well as between the nuclear magnetic moments and the magnetic field generated by the rotation of the molecule.

    Yeah, ok, that makes sense.

    Can't forget the Zeeman Effect

    Atheist, I did see a piece by one critic of "decoupling" that listed redefinition of the second as one possible solution to the so-called problem. He wasn't pushing it as a serious proposal, however. 

    I think it's way too early to attempt, however. We have only half a century of data directly comparing the caesium standard with observations. I'd stick with the current system of intermittent adjustments for at least a century or two, in order to get an ever more accurate measure of the average solar second. Hell, we could designate a nice round date in advance -- say, 2500 -- for when we'll switch to the "new second."

    It's not like we'll be messing with something that's cast in stone: the official UTC is determined by averaging the readings from dozens of atomic clocks around the globe, which differ by very tiny amounts depending on position. UTC had been in use for five years before we got around to introducing a correction so all read as if they were at mean sea level. (With altitude, even atomic clocks run slower due to relativity.)

    And throughout the 1960s (before UTC but after adoption of the caesium second), another correction was used to keep atomic time in sync with GMT. So fiddling with the definition of the second is not breaking totally new ground.

    Ah, for the good old days, when a meter was defined as the length of a metal bar in a Paris museum!

    I just curious whether the belief is that the cycles of a caesium 133 is actually consistent and stable or the variations are so "small" that we cannot observe them, like the theoritical strings of M theory.  And what impact do the other 8 dimensions of M theory have on time and space?

    Given the natural world with its non-linear and turbulent irregularity of the unfolding, it is interesting that it is built upon the stable reiteration like the cycles of the caesium 133 - or is that the most subtle of all variation in those cycles ripple out to create the patterns of weather and human consciousness.

    But, yeah, let restore the Sun-Earth standard as the definition of time.  Maybe Ron Paul will take it up as a companion cause with the Gold standard.

    Ron Paul? Gold standard? You calling me a Luddite, Trope? I'll let that pass.

    Yeah, I think the belief is that the atomic cycles being measured -- of various atoms, not just caesium -- are invariable. I've seen claims that the latest atomic clocks are accurate "within one second in a million years." Hard to prove or disprove; all we can say from the evidence is that -- once you factor out relativity effects -- atomic clocks do appear to stay remarkably in sync with other atomic clocks. Good enough.

    I think it's crucial that we retain the Sun-Earth standard as a constant reminder of the wobbly, precarious platform from which we humans observe the universe. The length of the day and the year have changed radically over time, the climate has gone from molten to ice-encased to something in-between, the atmosphere has only recently become oxygen-rich, and we only recently developed lungs to take advantage of all that delicious oxygen. We're blessed with a Goldilocks sun, but even it flickers and sputters, and one day will burn out, killing all our descendants if we haven't already gotten around to that by accident or design. The whole process is fragile, it's messy, it's in constant flux. We can observe the passing show with wonder and awe, and try not to add unduly to the chaos. Or we can try desperately to freeze-frame the moment, which seems pretty pointless.

    Now where was I? Oh yeah, about the proposal to ban the leap second ...

    So I get you're a wobbly, a Sun worshiper, one of those oxygen loving tree huggers, and a moral relativist to boot. Or I could have saved time & typing with "Canadian". 

    I'm just worried that if we give you an role in a relatively straightforward & instantaneous concept like "leap second", you'll drag it out into something painfully akin to curling.

    Listen, Peracles, I'm wobbly even by Canadian standards. But yeah, you pretty well nailed it. As for curling, I will happily sit down with a bottle of wine to watch a tournament, end to end, on TV. A superb, subtle and underappreciated game of strategy and skill.

    Never let it be said that Canadians don't have any stones. 


    I was just trying to make a joke with the Paul remark. 

    I do think you make a solid argument about going with the Sun-Earth standard.

    (and for what's it worth, I'm with you on the curling.)

    I took the Ron Paul remark in the spirit you intended, Trope. My gut feeling is that none of the delegates to this ITU meeting really thought through the can of bad-PR worms that this proposal would open. 

    The idea of dropping the leap second has been around for at least a decade, with similar arguments about simplifying and streamlining the functioning of navigation and communications systems. 

    But nobody has ever demonstrated the serious problems such a move is purportedly necessary to solve. After 40 years, it's fairly obvious that any software that might have been affected by insertion of a leap second has been successfully designed to accommodate the existing system.

    I suspect by 2015 the issue may not even be on the agenda. And yes, curling rocks.

    Latest Comments