Talk:Leap second

From Wikipedia, the free encyclopedia

WikiProject Time
Time Portal
This article is within the scope of WikiProject Time, an attempt to build a comprehensive and detailed guide to Time on Wikipedia. If you would like to participate, please join the project.
??? This article has not yet received a rating on the quality scale.
??? This article has not yet received a rating on the Project's importance scale.

Contents

[edit] Bias

Towards the end of the section titled "Proposal to redefine UTC and abolish leap seconds" there is a conclusion made that introduces a strong bias to the article, the purpose of the last part was to show firstly the cons of removing leap seconds and then the pros, but then an attempt is made to refute the pros in a manner of an argumentative piece which is not the purpose of an encyclopedia.

I propose removing the last part, as the same point is already made in the cons regardless.

[edit] Is there any practical considerations with regards to leap seconds?

Is there any practical considerations with regards to leap seconds? How does it affect anything? I suppose, for example, that the Unix epoch which counts seconds starting from January 1, 1970 would be about 22 (?) seconds out of sync with UTC, right? I think that discussions on such practical aspects needs to placed in the article. --seav 06:05, Jan 3, 2004 (UTC)

I believe it has something to do with GPS systems, airline computers and the like.... but i'm just drawing that from my head, from an article I read about the latest spin correction... I don't have enough info to add to the entry. Lyellin 06:12, Jan 3, 2004 (UTC)
Here someone asks a few relevant questions
* http://www.mail-archive.com/leapsecs@rom.usno.navy.mil/msg00051.html
and here is some related discussion, I think
* http://www.metrology.asn.au/leapseconds.htm
Essentially, there is a tension between our convention that a day is a fixed length, and our recording devices, because it isn't quite as we wish; I expect the tension will increase as the accuracy of our recording and measuring devices increase. However, I'm far from any kind of expert on the matter! :) Kyk 06:32, 3 Jan 2004 (UTC)
The following page has a very useful explanation of leap seconds:
* http://tycho.usno.navy.mil/leapsec.html

The only reason for leap seconds, in fact calendars vs timekeeping is to keep the Earth's surface position, and season timings in sync with the traditional points in the sky which mark the events. The reason for the gregorian calendar adjustment was to insert days into the mix to move easter back to occuring in it's traditional relation with the sky positions. (I don't recall these exactly and the details are not relevant to my point). The leap seconds are gradually realiging the earth's position back to day alignment as the leap year days adjust the calendar.

The earths nutation has caused the slipage and it was traditionally ignored, but there is still a significant amount to be accounted for, if I recall, it is a matter of inserting the adjustments to minimize other factors.


"Roughly 50000 years in the future, one can expect to have a day of 86401 seconds if the definition of the SI second is not eventually changed."

I'm not sure, but I think in the year 4000 the Gregorian Calendar won't work as it does now a days... so this phrase would be irrelevant... could someone check it out? I'm busy... --Henriquevicente 01:12, Apr 26, 2005 (UTC)

You may be falling into a (popular!) intellectual trap here, namely confusing the length of the day with the length of the year. The two are independent oscillations. Changing the leap-year rule near 4000 (no serious astronomer is confident to predict the exact length of the year that far into the future) would not affect the length of the day after that. --anon
The purpose of Leap Years in the Gregorian Calendar is to keep the calendar synchronized with the seasons, nothing more. The 4000-Year Rule as proposed by astronomer John Herschel, among others, i.e., dropping a leap year every 4000 years, is still just a proposal. Let 'em start worring about it, say, around 3990 A.D.! --QuicksilverT @ 18:58, 16 November 2005 (UTC)
It seems to me that if leap seconds are thought necessary because the earth's rotation slows slightly, it's the value of the second that requires adjustment, not the number of seconds in a particular minute. Smerdis of Tlön 21:04, 1 January 2006 (UTC)
Not sure what you're actually suggesting, but it's impractical to change the value of the second. Many very precise measurements are based on a constant value of the second, and you could cause chaos by trying to change that value. FireWorks 21:24, 1 January 2006 (UTC)
Well, I'm all for causing chaos; but it strikes me that some other unit of time with a different name needs to be adopted. The length of the day, the length of the year, and therefore the value of the second are not constants; they are astronomical facts beyond the reach of human standards. Smerdis of Tlön 23:29, 1 January 2006 (UTC)
I think you'll probly find that there's two seconds: One of which is 1/24*60*60 of a day long, and one of which is precisely "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", whatever that means. It's a convenient fiction that they're the same, when in reality they've never been the same (at least as long as the latter's been defined), thus quoth: [1]. Because it's easier to make international standards based on the second definition (apparently) our second count is based on that. I think it would've been better if the SI had've called the latter one the "newcomb" or some such. It's probably relevant that before 1972 the basis for civil time did use these variable-length seconds. —Felix the Cassowary 14:05, 2 January 2006 (UTC)

[edit] Frequency of changes

The article states that "Historically, leap seconds have been inserted about every 18 months". I did a quick check of the table and it looks like the interval has been 18 months only 5 times, while it has been 12 months 14 times. Might it be better to start that paragraph at "The Earth's rotation rate is unpredictable..."?

Looking at the frequency of updates and the fact that we have gone 7 years without one, the higher frequency up to 1998 would appear to be some sort of gradual 'catch up'. Would that be a correct interpretation?

82.43.52.87 16:16, 30 July 2005 (UTC)

The frequency of about every 18 months is correct for all leap seconds between 1 January 1972 and 31 December 2005, including the initial 6 month interval and the final 84 month interval, because the mathematical average during that period was 17.7 months. A graph showing the relationship of UTC to UT1 can be found at [2], showing that there was no attempt to 'catch up'. Rather, Earth's rate of rotation actually sped up slightly during 1997. — Joe Kress 21:15, 30 July 2005 (UTC)

I quite liked this: http://www.post-gazette.com/pg/05210/545823.stm although I have no idea if it's suitable as an external link on the article.

I also like it and I'll add it. — Joe Kress 03:54, 31 July 2005 (UTC)

[edit] Simple naming question

Shouldn't this be called a leap day? Like... a leap year has an extra day, a leap day has an extra second? :-P --67.172.99.160 23:51, 5 August 2005 (UTC)

Interesting question. Unfortunately, "leap day" is already the name for the extra day in a leap year. There was no doubt some discussion in the years leading up to their adoption in 1972 concerning what to call them. We only know the result of the discussion—the ITU decided that they will be called leap seconds. — Joe Kress 03:31, 6 August 2005 (UTC)
On that basis it would be more precise to call it a "leap minute" because the effect is to increase or decrease the length of the last minute in the day: 23:59:60 is the extra second added for a positive leap second, 23:59:59 would be skipped for a negative leap second. EdDavies 23:26, 2 January 2007 (UTC)

[edit] Removed incorrect information

The reason we need leap seconds is that the rotation of the Earth is slowing down. The solar day does gradually become longer by about 1.7 ms every century, mainly due to tidal acceleration from the Moon. The SI second that is counted by atomic time standards has been defined in such a way that its length did match the nominal second of 1/86400 of a mean solar day some time during the 19th century. Since that time the length of the solar day has been slowly increasing. Therefore the time as measured by the rotation of the Earth has been accumulating a delay with respect to atomic time standards. Whenever the accumulated delay approaches one second, a leap second is added to UTC.

This misconception weas contradicted in the paragraph that followed it. Ben Arnold 07:05, 8 August 2005 (UTC)

The paragraph you removed is completely correct. So is the paragraph which follows it in its own way. What seems to be missing is an obvious linkage between the apparently slow rate at which Earth is slowing down (milliseconds per day per century) and the apparently much more rapid rate at which leap seconds are added (many seconds per century). That is immediately achieved by restating the rate at which Earth's rotation is slowing down, 1.7 milliseconds/day/century (actually 1.7 ms/(d·cy)), in square centuries. To do this we multiply by the number of days in a century, which is 36525 days per Julian century (to two significant digits, it does not matter what kind of century is used, but astronomical equations are always stated in terms of Julian centuries, never in terms of Gregorian centuries or mean tropical centuries). That yields a rate of 62 seconds/century², which is obviously the proper magnitude to yield many leap seconds per century. But the actual mathematical equations should be given, which requires integration. I am composing a rewrite. — Joe Kress 17:40, 8 August 2005 (UTC)


Actually, it is not correct that the reason we need leap seconds is because the Earth is slowing down. We need leap seconds because the SI (atomic) second is not exactly 1/86400 of a mean solar day. Even if the Earth's speed could be locked today, and never changed again, leap seconds would still be needed at about the same rate we're used to. It's the difference in the two units, not the rate of change that matters.

Besides, it's not really that we "need" them, so much as that we have decided to use them. When UTC was established in 1972, harmonizing the definition of the civil second and the atomic second, we chose to adjust for the inevitable differences between the rock-solid UTC and the highly variable UT1 (earth rotation time) in one-second jumps.

Prior to that, the adjustments were still being done, but it was done as a fractional disseminated-frequency offset from the atomic time definition. That is, an adjustment trim was applied to the disseminated GMT time and frequency standards in order to keep them in fairly close agreement to actual earth-rotation time.

Then, by '72, it was decided to scrap GMT, and quit trying to keep adjusting it for minor earth-rotation variations. UTC was established, the disseminated frequency offset was set to zero, and, since then, civil and atomic clocks have ticked in sync. Standards organizations stopped trying to adjust the time scales, and let them drift apart, but published the predicted and measured differences between UTC and UT1, for those interests that required earth-rotation time to a precision of less than one second. Only when the differences threatened to accumulate to more than 700 ms would an ajustment to UTC be scheduled. These are the familiar leap seconds, scheduled (or waived) six months in advance for 23:59:60 UTC on the last day of June or December. --Jeepien 05:03:27, 2005-08-09 (UTC)

If we do not explain that the Earth is actually slowing down, then it appears as if the SI second was set at the wrong value to stay in sync with mean solar time. So it would appear as if we would forever be fudging our clocks with leap seconds because of an initial mistake; it would be then more sensible to re-define the SI second. However, this is is not feasible because the Earth WILL be getting out of sync with the SI second because it is actually slowing down irregularly. So this information is totally relevant when explaining why we have leap seconds. -- Tom Peters 10 August 11:20 UTC


In the very least, this section is confusing. I had to find other sources in order to decipher it. It should in essence read "The reason that leap seconds are needed is because the SI second is not exactly equal to a solar second. The reason for this difference is that the SI second was defined in the 19th century and due to the slow change in the earth's rotation the SI second now differs from the solar second by X. While, X is very small, we are talking about a second and when you consider that there are 86400 seconds in a day and 365 days in a year this small difference is multiplied many times and results in a whole second of difference in only a matter of years. Basically, the SI second is based on an absolute amount of time that corresponded to a an average second in the 19'th century but is now off by X."

I would also like to see some reason why this was adopted rather than some "Atomic Equation of Time" that one could use to translate between atomic and solar (or sidereal) time. Or at least the existence of such. Does such a method exist now, for something like astonomy? Syscore 20:32, 18 August 2007 (UTC)

[edit] The Earth is... umm... speeding up.

All of this discussion about how the earth is slowing down ignores one inconvenient fact. The earth is, at present, not slowing down. It is, in fact, speeding up, and has been doing so since 1972, coincidentally just around the time that UTC was introduced.

When the International Atomic Time (TAI) scale was zeroed to civil UT, on 1958-01-01, the mean solar day was about 1.4 ms too long (as compared to a standard SI day of 86 400 s). Over the succeeding 14 years, the equivalent of around 10 s of "leap" time was introduced between TAI and UT, but in those days the adjustments didn't "leap". They were done by slightly padding civil time using a flexible frequency offset from TAI, so the adjustment was made continually, on the fly. Furthermore, in those days, the earth actually was slowing down. Over the course of that same 14 years, rotation slowed until it reached more than +3.1 ms/day relative to the SI day.

On 1972-01-01, when UTC was adopted, and set to 10 s offset from TAI exactly, it proceeded to tick at the same rate as TAI, i.e., 1 UTC second = 1 SI second, exactly. Since then, 22 leap seconds have been added, with one planned for 2005-12-31, which will bring the total offset between UTC and TAI to (TAI - UTC) = 33 s.

However, for the last 33 years something else has been going on. The day has been getting shorter. For some reason, no doubt related to internal fluid mechanics, the Earth's crust has been accelerating. By the time the last leap second was introduced at the end of 1998, the day length had shrunk back to something like +1.3 ms/day. It is this continued acceleration that accounts for the fact that leap seconds, once a circannual phenomenon, have become relatively rare. This winter's leap second will be the first in seven years.

During that seven years the acceleration has continued. The difference between a mean solar day and an SI day has now essentially vanished. Over the most recent 48 solar days for which data is available, 22 of them have been shorter than an SI day! If this acceleration continues much longer, at some point in the future a negative leap-second may be needed.

So it is not correct to blame leap seconds on tidal deceleration. In the first place, ever since they were introduced, there hasn't been any. Or, more correctly, the small tidal deceleration that exists has been swamped by the much larger short-term earth-rotation variability. And this the rule, rather than the exception. The vertical plate movement responsible for the "Christmas Tsunami" of 2004 added measurably to the angular velocity of the Earth's crust. No doubt other geophysical events will have different effects.

Although the current period of acceleration probably can't be maintained for many more years, and while the +1.7 ms/cy tidal effects (which some sources suggest may really be closer to +1.4 ms/cy) will continue to slow the planet over geological time scales, these effects will aways be swallowed up in the "noise" of short-term variations on the order of a human lifespan. --Jeepien 07:19:41, 2005-08-11 (UTC)

My analysis produces somewhat different results than yours. I used the annual change in ΔT, which is TT − UT1, where TT (Terrestrial Time) is TAI + 32.184 s, close to the average time during the nineteenth century, which defines the SI day. For the last three years (2003.0-2005.0) I calculated ΔT from various IERS bulletins B (ΔT = 32.184 s + (TAI − UTC) − (UT1 − UTC)). I find that ΔT increased by 3.3 ms/d for 1898-1919, 1.2 ms/d for 1920-1927, −0.1 ms/d for 1928-1939, 1.2 ms/d for 1940-1964, 2.6 ms/d for 1965-1984, 1.4 ms/d for 1985-1990, 2.1 ms/d for 1991-1998, and 0.6 ms/d for 1999-2005. Near 1972, Earth's rotation rate actually slowed even more than it had before! Only during the last seven years (and during 1928-1939) can it be said to have sped up. Of course, the rates did vary a little within each period and for years at the limits of each period.
Although the SI second and day is usually stated to be based on and hence defined as the average day between 1750 and 1890, the average day for the period 1686-1823 was the SI day (with a change in ΔT of less than 0.01 ms/d), ΔT changed by −0.5 ms/d for 1824-1834, 0.1 ms/d for 1835-1861, −0.5 ms/d for 1862-1866, −1.0 ms/d for 1867-1874, and −0.1 ms/d for 1875-1897; or an average change of −0.08 ms/d for the entire period of 1686-1897 (over two centuries). Of course, the long term (over 2700 years) lengthening of the mean solar day is about (1.7 ms/d/cy)T, which produces a parabolic ΔT of about (31 s/cy²)T², so the mean solar day was actually lengthening slightly during the base period. What source places the rate closer to 1.4 ms/d/cy? — Joe Kress 18:23, August 12, 2005 (UTC)
I don't dispute your figures, but I think if you look at the numbers graphically, it will leap out at you. Here's a graph of excess length of day (LOD) from around 1720 until 2003. Positive slopes correspond to periods of deceleration, negative slopes to acceleration. The values are measured directly in ms, and correspond to the difference between the observed day length and a standard (86 400 s) SI day.
As you can see, the peak near 1972 is the highest (i.e., slowest) in a lifetime, and since then, but for backsliding during the 1990s, the general trend has been downward (faster). If the last two years were shown, a smoothed average would be very close to zero.
As long as the LOD difference hovers around 0, leap seconds are rare, and if it manages to drop below zero for any length of time, negative leap seconds will be required. Nobody is predicting that it will, but then again, nobody is clear on what's causing this current period of acceleration, either. --Jeepien 00:00:35, 2005-08-13 (UTC)

[edit] Proposed change

I am proposing replacing the entire section on Reasons for leap seconds with the following. Note that this would remove the reference to Creationists. I don't understand what that issue is supposed to be, and it wouldn't belong here anyway. Sure the earth was going faster in the past, so what? Any comments on the following text would be appreciated. --Jeepien 21:35:17, 2005-08-11 (UTC)


For most of history, the measurement of time has been an exercise in astronomy. Traditionally, the second was defined as 1/86400 of the length of a mean solar day. Units of time depended on the Earth's speed of rotation on its axis and the properties of its orbit around the Sun. For any ordinary purposes, this was fine but, as clock-making technology improved, the problems with using Earth as the standard timepiece became more evident. With the invention of the quartz clock in the 1930s, the best available timepieces were becoming increasingly stable. The length of a solar day, even when averaged over a year, proved to be anything but. Factors such as winds, ocean currents, plate tectonics, glacial melting, and fluid currents within the Earth's interior all combine to add a complex set of wobbles to the Earth's rotation, speeding it up or slowing it down slightly in unpredictable ways, often for decades at a time. Over the long term, however, the Earth is slowing down. The tidal effects of gravity between the Earth and the Moon cause the mean solar day to get longer by approximately 0.0017 seconds each century, on average. Compared to the short-term speed fluctuations, this is a small effect, but it is constant and inexorable, and will become more significant over time. All of these factors make the Earth a sub-standard timepiece.

Whenever the tools of measurement become more precise than the standard of measurement, a new standard is needed. With the development of the caesium atomic clock in 1955, an extremely precise and stable time scale became available. On 1958-01-01 the International Atomic Time (TAI) scale was defined, and set to match the civil time scale on that date. At first, TAI was used primarily by scientists, but proved itself to be a superior tool for all timekeeping. In 1967, this new definition of the second, based upon the vibrations of caesium atoms, was adopted as the official SI unit of time. Planning was begun to change the civil time scales around the world from the old astronomical standard (commonly called GMT), to the atomic standard. The switch was officially made on 1972-01-01. However, TAI could not be used directly; since 1958, mean solar time, which was still tracking the wobbly Earth, had drifted about 10 seconds "slow" with respect to TAI. A new scale, Universal Coordinated Time (UTC) was created, and simply set equal to TAI - 10 s. The two scales would thereafter remain locked at the same rate. Still, it was clear that more adjustments would be needed to keep UTC and mean solar time from drifting apart again. It was decided that future adjustments would be made in precise one-second steps. These steps, called leap seconds, allow the time and duration of each second to remain locked to the atomic standard, while making sure that the average time that the Sun crosses the Greenwich meridian is still noon, give or take 0.9 s. No such adjustments are ever made to TAI.

Since the inception of UTC, there have been 22 leap second adjustments (see list), all of them "positive", i.e., adding an extra second to UTC as opposed to skipping one. Leap seconds are announced six months in advance, and occur simultaneously around the world during the last minute of June or December, at 23:59:60 UTC. In recent decades, the Earth has been in a period of acceleration, so fewer leap seconds have been needed. The one announced for 2005-12-31 will be the first in seven years, bringing the difference between TAI and UTC to 33 seconds. That is, UTC-TAI = -33 s as of 2006-01-01.

My comments:
  • somewhat too extensive: it is a (brief) account of the history of the second, which should be or already is described in other articles: use links.
    • It is, I believe, shorter than what's there now, isn't it? I would also reduce the now-redundant info in the Announcing section, or move the relevant info there.
  • tidal deceleration contributes +2.4 ms/cy to the l.o.d.; the number +1.7 is the observed average over the past 25 centuries. The difference is probably mainly due to so-called glacial rebound, which is another long-term process since the end of the ice age; see the article on Delta-T.
    • Tidal effects should contribute that much in theory. This has never been observed.
      • Only the total effect of all processes can be observed. The partial contribution of known mechanisms, like the tidal deceleration, can be accurately modeled; I can give you scientific literature references. They are consistent with the OBSERVED acceleration of the Moon. The laws of preservation of energy and angular momentum then allow to compute the effect on the rotation rate on the Earth of the tidal effect by itself. There is a long-term discrepancy of about 0.7 ms/cy with the observed change in rotation rate, which is due to other long-term mechanisms. Models of glacial rebound largely explain this effect. So IMNSHO it is totally OK to mention these facts.
  • Jeepien, on your previous comment: I begin to find the discussion meaningless. The reason that we need leap seconds is that the rotation of the Earth is irregular, but nonetheless for civil life we want to keep in sync with the solar day. There are a zillion mechanisms that influence the rotation rate on all time scales, but the tidal deceleration is the main and most persistent one. I don't understand why you and some others are so opposed to even mentioning it. In any case I object to giving so much attention to recent short-term irregularities.
    • I don't recommend leaving it out, and in fact I include it and mention that by its inexorable nature it can't help but be significant in the long term. But the fact remains that short term effects are an order of magnitude more significant than tidal effects in any given century.
      • The rotation rate of the Earth has the statistics of a random walk: sometimes it is accelerating, sometimes it is decelerating. There is a long-term trend, driven by tidal deceleration and probably glacial rebound. Seasonal effects (winds, ocean currents) work both ways but do not exceed a few ms (accumulated). There are some longer-term (decades) mechanisms that can have an accumulated effect that will require leap seconds, but these can work both ways too: redistribution of mass between poles and aequator (ice caps) and exchange of angular momentum between core and mantle.
  • I did not add the piece about creationists misinterpreting the reason for leap seconds, but apparently this is an issue and if it should be addressed, this is the proper place; possibly under its own sub-heading. I say we keep it.
    • This one is a no-brainer. Creationism deserves no mention in any scientific article. If it's an issue, deal with it under Creationism where it belongs. It is certainly not an issue to the community of earth-rotation interests. But I would be grateful to anyone who could explain what that paragraph says. It seems to say that leap seconds should not be confused with amount that the earth has slowed, yet that's exactly what they are.
      • As I explain on the page, the creationists are confusing rate (velocity) with time passed (distance traveled). So they believe that whenever a leap second is counted, that the length of day has increased by 1 second, e.g. that the solar day has increased from 86400 to 86401 seconds. Counting backwards that way, the Earth would have had a ridiculous fast rotation rate in the recent past; which they take as proof that scientific chronology is false and (non sequitur) that the biblical short chronology is true. In any case, this distinction apparently is so confusing that it is proper to explain it here, even without creationist discussion. The anonymous anti-creationist tried to explain the difference between rate and time passed with an example, I elaborated on the concepts. If this still appears unclear to you, please try explain yourself. Iterum censeo that iff there is a vocal party that tells nonsense about the topic (leap seconds) then the page dealing with that topic is the place to refute it.

-- Tom Peters 20050812T10:15 UT

  • It's this kind of thing that shows people just how much credibility we really lack. Why is it necessary to attack the beliefs of creationists in this article? What did the article have to do with the beliefs of creationists in the first place? Also, the page dealing with a particular topic of issue is not the place to refute it, but rather, that article's corresponding Talk page. In closing, I believe a rewording of all parts of the paragraph, in addition to the removal of any reference to creationists (which separates the wrong and right interpretations of the need for leap seconds, making it that much harder to understand), and possibly merging it with the paragraph immediately following, would make the article easier to understand for most people (including myself).--JEmfinger 04:49, 23 December 2005 (UTC)

[edit] Leap hours predetermined?

With the proposed leap hour methods, are the leap hours predetermined like leap days and unlike leap seconds? Unless they are, the arguments in favor of leap hours seem completely bogus: you would still need to consult a table for leap hours, and you can still not predict time far into the future (albeit you can on a much larger time frame).

  • This is of course assuming that the human race hasn't destroyed itself by the time solar time and UTC differ by 3600s. Gee Eight 22 December 2005 21.12 UTC

[edit] Leap seconds on 31 December?

Am I the only one who has noticed that the introduction of a leap second at 23.59'60" UTC on 31 December is a bit unwitting, as that is the moment when quite a lot of people will be counting down official UTC seconds to the new year? Why and how will I, being British and therefore a UTC timezone resident, still be able to count down from 10 at 23.59'50" with Big Ben next Saturday despite the impending 2005 leap second threatening to make me miscount? Can someone please tell me where my logic has failed (preferably before the 31st) as this is really confusing me! Gee Eight, 22 December 2005 21.01 (unless they've added the leap second early) UTC

  • Your logic hasn't failed, but unfortunately, there's no way of getting around this problem as far as I can tell. The only thing I can really advise is that, when there's 10 seconds remaining, start counting down from 11! --JEmfinger 05:31, 23 December 2005 (UTC)

Or they could just start the countdown at 23.59'51". I'll synchronise my watch with UTC later today and see. (Interesting point: the man responsible for keeping Big Ben on UTC has within the last 36 hours been given an MBE...is something going on here?) Gee Eight, 31 December 2005 15.12 UTC

Or, to be more accurate (as I understand it) with the idea of a repeating second (at least in some timescales), they could countdown "10...9...8...7...6...5...4...3...2...1...1...'Happy New Year!'". Of course, I guess it matters which timescale you're using, and I think UTC inserts a second rather than repeating one, so just adjusting the start time of your countdown would probably be more correct for the typical human-used timescale. OK...this didn't really add much to the discussion, but seemed slightly humorous, so I figured I would add it. Igjeff 16:59, 31 December 2005 (UTC)

What the BBC actually did on their countdown on News 24 that night was to repeat the 2 second count, so it went 10...9...8...7...6...5...4...3...2...2...1...0. It's viewable on YouTube, though apparently the countdown projected on Canary Wharf, shown in the -40s, had already been adjusted. -- Arwel (talk) 19:49, 14 January 2007 (UTC)

[edit] i think i spotted a vandal

someone who knows the truth check the edit by (17:00, 1 December 2005 70.23.27.61) where he/she changed a single date from 1890 to 1892. The edit is in the sentance: "The SI second that is counted by atomic time standards has been defined in such a way that its length matched the nominal second of 1/86400 of a mean solar day between 1750 and 1892." It is in the "Reason for..." category. There was no reason given why the date was changed in the first place. I've noticed that vandals try to change dates like this because they are hard to spot. 71.131.52.62 06:28, 26 December 2005 (UTC)

The change is appropriate. 1890 is a rounded version for 1892. F. R. Stephenson, IIRC, used the range 1750-1890 when he selected 1820 as the vertex of his parabolic representation of Delta T, whereas the actual observations used by Simon Newcomb, on which he based his tables, which form the basis of Ephemeris Time, end in 1892, not 1890. — Joe Kress 07:54, 26 December 2005 (UTC)

[edit] Spelling errors

Corrected spelling errors in article. (IchBin 05:46, 1 January 2006 (UTC))

[edit] BTW, did you notice...

That 2005-12-31 T 23:59:60 Z was the first leap second observed since the founding of Wikipedia? --  Denelson83  08:21, 1 January 2006 (UTC)


[edit] Example

After exactly 500 rotations, your counter will register 43,200,001 seconds. Since 86400 × 500 is 43,200,000 seconds, you will calculate that the date is 12:00:01AM on May 16, 1971 (exactly 500 days after January 1, 1970) as measured in atomic time (UTC), while it is only 12:00:00AM on May 16, 1971 in solar time (UT1). If you had added a leap second on December 31, 1970 to your counter, then the counter would have a value of 43,200,001 seconds at midnight on May 16, 1971 and allow you to calculate the correct date. The actual system involving leap seconds was set up to allow TAI and UT1 to have an offset of 0 seconds on January 1, 1958.

This is exceedingly confusing. Could someone who understands it please reword it? It says that your counter will read 43,200,001 on May 16, but if you add a leap second, then you get 43,200,001 on May 16 and can calculate the actual date. The impression that that sentence gives is that you'll get 43,200,002 on your counter, and then you just have to fudge everything up to get the real date. Obviously this wasn't the intent of the writer... FireWorks 21:40, 1 January 2006 (UTC)

[edit] crazy approach

i'm just a guy who likes math, and knows nothing about leap second. what i do know is that the problem happens because there's one count of the time from Earth's translation movement and another one counting atoms. so that makes a difference of 1 second almost every 6 months between them.

as long as we are made of atoms, and we are moving around the Sun, i think a better approach would possibily measure another kind of frequency. once again, i'm no expert in those measurements details, but i do know that, logically, it makes more sense for us, as humans, measure frequencies closer to us, and that could potentially avoid any kind of "leap", not only in 6 months, but even in centuries.

my suggestion would be looking for something around carbon 12. either diamonds or maybe plain old coal. maybe, just maybe, a carbohidrate. the simpliest the better. now, that may sound pretty stupid for most specialists, but try to get the big picture. why did we started to use any of the ways we do today? because it was experienced they're stable at our home, some parts of earth's surface. it's already proved even that's untrue, on time dilation. so, why bother trying to find something that's whatever ns more precise? better if we can find a way to precisely measure things that walk in time in same "speed" than we do. that way it will keep changing over time, but it will be same than human.

maybe measuring the eletrons moving in water molecule. maybe measuring light (or photons) going across a carbohidrate. i'm really not sure which methods could be used, but i'm sure it's not impossible to find a one, and i believe it would be the more appropriate the closer it gets to our form of life.

--Caue (T | C) 13:36, Thursday March 30, 2006 (UTC)


Er I don't think you quite understand the issues here. Trying reading Atomic clock and Delta T and perhaps a bit more on time. If we really want to measure solar time, sundials are your best bet... Using the frequencies of other atoms is going to give the same result (albeit less accurately if at all possible) It's nothing to do with time dilation. We are moving in TAI (atomic time/realtime) more or less. Solar time simply depends on how fast we're going round the sound (more or less). A second is a second. It wouldn't be particularly smart if the length of second varies. Solar time is important but it isn't the end all. Nil Einne 17:00, 5 November 2006 (UTC)


Its not an issue of accuracy, it's one of standards. The "atoms" did walk in time the same way we did in the 19th century. However we are just a tad slower now and since the SI second is still defined as the same fixed number of oscillations the SI second is a bit faster now. Multiply that error in the second by 86400 seconds in a day and 365 days in a year and you get a whopping one second error in only a few years. What would seem to make sense to me is that you do the math ahead of time and publish real solar seconds. There will always be leap seconds (or math) involved since you are comparing a fixed time reference (atoms) to a changing time reference (the earth). SI seconds are needed because they are an absolute reference and when I say that some phenomena occurs in X seconds I mean X SI seconds regardless of when I said it. They are problematic though when trying to match them to what we think of in terms of calendar seconds, days and years.Syscore 20:52, 18 August 2007 (UTC)

[edit] merge with Coordinated Universal Time

Leap seconds are solely a feature of UTC, and UTC cannot be defined without discussion of leap seconds. They form a single topic, so it is silly for them to have separate articles. 81.168.80.170 12:10, 29 April 2006 (UTC)

Both of these are fairly long articles. They seem to stand well on their own, connected by links. And after all, leap seconds are a relatively small part of UTC, so either information would be lost doing the merger or leap seconds would get a disproportionate amount of space in the combined article. I favor leaving them the way they are.
--Rbraunwa 21:28, 28 May 2006 (UTC)

I agree with Rbraunwa. I know nothing (well, a little now) about UTC, I found this article looking for information on leap seconds. If they did get merged, make leap second a redirect but this article can stand alone. Rahulchandra 09:25, 8 June 2006 (UTC)

[edit] Encyclopedic tone

The encyclopedic tone of this article could be improved through the removal of personal phrasing, such as repeated use of the word "you" and phrases like "Take care not to confuse the difference between..." Robert K S 14:44, 18 May 2007 (UTC)


What would you write in place of these types of phrases? I to am in favor of encyclopedic tone but I guess I never thought engaging the reader was non-encyclopedic. Just curious. Maybe I should look this up.Syscore 20:55, 18 August 2007 (UTC)

[edit] collapsing table, a bit

Since there was no leap second from '99-'04, I'm going to shorten the table to have the box read:
1999(br)through/thru(br)2004|none . (Through is too wide). If 2008 doesn't get a leap second, that can be collapsed as well. It should have a span of at least four years, as the 'thru box' takes up three and hence takes up the same amount of space. I'm also shortening 1973 through 1979, as the results are the same. If anybody is against this and decides to revert it to the expanded version, please let me know that you will/did. The reason I'm against the expanded one is that it is getting lengthy as it's 37 rows long. A shortened version will have 31 rows if I add a "Year|June|December" box at the bottom so people won't have to go to the top if they do happen to forget which is June and which is December. Socby19 19:05, 29 October 2007 (UTC)

[edit] Passive voice

This article is suffering from an overload of passive voice. I came here specifically to find out what standards body announces the leap seconds, and I encountered an infuriating series of passive-voice statements that seem determined to keep this information secret:

"...UTC is occasionally corrected by an intercalary adjustment..."
"UTC ... is kept approximately in sync with UT1 (mean solar time) by introducing a leap second when necessary. This happens when the difference UT1−UTC approaches 0.9 seconds, and is typically scheduled either at the end of June 30 or December 31 though leap seconds can be applied at the end of any month."
"The announcement to insert a leap second is usually issued whenever..."

Can someone who knows please change the article to say who adds the leap seconds? --P3d0 (talk) 14:37, 21 November 2007 (UTC)

It's in Leap second#Announcement of leap seconds. The International Earth Rotation and Reference Systems Service (IERS) announces leap seconds in IERS Bulletin C whenever they predict that the difference UT1−UTC might exceed 0.9 seconds at either the next January 1 or July 1, both with and without a leap second. Also see Earth Orientation Center and click on "leap second" at the upper left. I've added IERS to the introductory paragraph. — Joe Kress (talk) 22:25, 21 November 2007 (UTC)

[edit] US Law

The page currently has "Law of the United States indicates that the legal time of the US is based on mean solar time."

I believe that US legal time has now been legislated to be UTC-based. Perhaps an American can check and if appropriate edit the page.

82.163.24.100 (talk) 13:38, 23 November 2007 (UTC)

Your are correct. I'm removing that sentence, which is no longer germane. — Joe Kress (talk) 21:26, 23 November 2007 (UTC)

[edit] Announcement of leap seconds

BIPM's most recent Bulletin, like many of its predecessors, states

"NO positive leap second will be introduced ..."

I point out that this wording leaves the introduction of a negative leap second completely undefined.

Perhaps there is a more rigorous French version?

82.163.24.100 (talk) 18:42, 24 December 2007 (UTC)

No French version appears to exist. Even the French version of this Wikipedia article, Seconde intercalaire, only provides a link to the English Bulletin C. Of course, because the English version is provided by the French themselves, I trust their own translation. However, I did find a French article from l'Observatorie de Paris announcing the 2005 leap second: Une seconde de plus, including the variation from the long term trend of Earth's decreasing rate of rotation (the long term trend has been removed). Another graphic from the IERS shows that its rate of decrease has only slowed. Only if it increased would a negative leap second be needed. — Joe Kress (talk) 00:22, 25 December 2007 (UTC)

[edit] Source for real rate of change in solar (or is it earth?) day

The actual source for the 1.4 ms per day increase in day length is quite different from the one stated in the NASA web page (2.3 ms per day). As both may be considered trustworthy (although I would prefer NASA as a reference), we have to choose or seek other references to make this info accurate (I think it is somewhat essencial to the article). Preymond (talk) 15:56, 2 April 2008 (UTC)

I'm not sure where the USNO got their figure of 1.4 ms/cy, but it disagrees with other sources. Some of this could be due to the period over which the rate applies. NASA's figure of 2.3 ms/cy is only one aspect. As stated in ΔT, the rate due to tidal friction alone is 2.3 ms/cy, but glacial rebound since the last ice age reduces that by 0.6 ms/cy to 1.7 ms/cy as the average rate over the last 27 centuries. All three of these figures come from Stephenson and Morrison as cited in ΔT, who personally studied the ancient eclipses. They note that the rate does fluctuate over periods of several centuries. — Joe Kress (talk) 16:40, 2 April 2008 (UTC)