Talk:Leap second

Using a 2018 clock to tell time in 2010?

The following text was added to the article recently: “Indeed, if an archive indicates a time stamp of 2010-10-01 00:00:00 UTC, this time was 2010-09-30 23:59:57 on a 2018 clock, since in the meantime three leap seconds were inserted.” This makes no sense to me. There is no such thing as a "2018 clock" there is only the UTC clock. Even if there were a "2018 clock", how could you use it to measure time in 2010? This sentence needs a better explanation for it to make sense, or it should be deleted. John Sauter (talk) 21:02, 14 October 2018 (UTC)

I understand what problem the sentence is intended to describe, but I agree that we need a clearer way to state it. Perhaps a mention of Unix time? Dbfirs 21:24, 14 October 2018 (UTC)

I hope there is no need to mention UNIX time. UNIX time is poorly designed: it only covers 1970 to 2038 on 32-bit systems, and it does not have a unique representation for a leap second. Perhaps the meaning is “If you use a clock which does not include leap seconds, and it is synchronized to UTC in 2018, if you extend that clock back to 2010 it will not correspond to UTC because of the omission of leap seconds.” If that is the intended meaning, the sentence adds nothing useful to the article. Having seen no objection, I am removing the sentence. John Sauter (talk) 21:19, 20 October 2018 (UTC)

I agree with your removal of the text. Dbfirs 06:45, 21 October 2018 (UTC)

Unix time isn't poorly designed, it's poorly and lazily implemented. From the article - "since 1970-01-01 00:00:00 in Unix machines or since 1900-01-01 00:00:00 in NTP. This counter has no indicator that a leap second has been inserted..." That statement, and the related POSIX definitions, are self-contradictory. If Unix/POSIX (even NTP) time actually counted seconds in an epoch, the leap seconds would be in there. Fact is, they don't. They simply follow the incorrect assumption that there are no leap seconds, so when a leap second occurs they become discontiguous - they repeat or delete a second, just to make conversion between their so-called "epoch time" and wall clock time easy to do, since accounting for leap seconds is then unnecessary. Current Unix time has 27 fewer seconds than there have actually been since 1/1/70. Msauve (talk) 23:16, 22 December 2019 (UTC)

I consider the POSIX definition of time_t as “Seconds since the Epoch” and simultaneously as an encoding of UTC as the design, and the code that attempts to implement that definition as the implementation. On that basis, the problem is with the design. I agree that the design is poor, but I would hesitate to call it lazy, since I don't know what constraints the authors of that part of POSIX were under. John Sauter (talk) 00:21, 23 December 2019 (UTC)

What? No, the minimum representable date is Friday 1901-12-13 (not 1970), and the maximum representable date is Tuesday 2038-01-19 in the 32 bit Unix time. I will remind you that the most significant bit is actually a sign and when it is 1, it will be negative (in this case less than 1970). 2A00:1FA0:42F5:C6EF:9152:8171:73ED:3C42 (talk) 15:23, 12 June 2020 (UTC)

Where in POSIX does it specify that time_t is signed? In the Single UNIX Specification sys/types.h page, it merely says that "time_t shall be an integer type.", not that it shall be a signed integer type.

In practice, few if any UN*Xes have it as an unsigned type, but "the POSIX definition of time_t" is "what the Single UNIX Specification says", not "what UN*Xes do in practice, over and above what the SUS requires". Guy Harris (talk) 16:01, 12 June 2020 (UTC)

Can leap seconds really be other months?

It seems the IERS doesn't think so: "Leap seconds can be introduced in UTC at the end of the months of December or June, depending on the evolution of UT1-TAI." — Preceding unsigned comment added by 128.240.225.21 (talk) 11:02, 10 October 2019 (UTC)

Yes, up to the present, only the months of June and December have ever been used for leap seconds, and these will always be the preferred months, with second choices being March and September, but the standard allows any month to be used if a catastrophic volcanic eruption or earthquake suddenly changes the length of the day. The overarching aim is to keep Coordinated Universal Time within nine-tenths of a second of Universal Time. I haven't mentioned GMT because some sources use it to mean UTC and some to mean UT1. Dbfirs 20:08, 10 October 2019 (UTC)

Plan to delete "Questioning the utility of leap seconds" and replace with (by moving up) "International proposals for elimination of leap seconds"

Although I have deleted the most blatant OR and opinionising [is that a word?], the sub-section "Questioning the utility of leap seconds" is unsourced and IMO, adds nothing to the article. The sentiments it expresses are given far more succinctly (with citations!) in "International proposals for elimination of leap seconds". It seems to me that it would make a lot more sense to simply delete it and lead the "Future of the leap second" section with the International proposals subsection. As I have already wielded the pruning saw fairly heavily, maybe it is best if I pause for comments? --John Maynard Friedman (talk) 17:24, 27 October 2019 (UTC)

I believe I improved this content by removing heading and the first sentence. I am not opposed to further improvement. It seems good to describe problems associated with leap seconds before we get to the proposals to eliminate them. ~Kvng (talk) 14:22, 30 October 2019 (UTC)

Yes, that was a big improvement and made the article a lot more readable. I've gone further, deleting some more trivia from that new "sub-lead" so that it really concentrates on stating the essence of the problem and what needs to be done about it. I've moved the International Proposals up to follow that new sub-lead because I believe that most readers want to know what is being done about it before getting into details on the practical issues arising from each time a leap second is created. Looking forward, I know that Wikipedia is not a guide but I do think it would be good to work towards providing an information source that forewarns readers about what they may need to do to get ready. --John Maynard Friedman (talk) 18:37, 30 October 2019 (UTC)

Proposed new version of the lead

IMO, the lead as it stands at present doesn't meet the standards of WP:LEAD. It doesn't provide a succinct summary of the article and gets bogged down in detail. I have drafted this replacement text. As it is a significant change, I thought I should check first whether anyone disagrees. (Some of the details that I weeded out will need to be relocated, probably at history.

A leap second is a one-second adjustment that is occasionally applied to Coordinated Universal Time (UTC), to accommodate the difference between precise time (as measured by atomic clocks) and imprecise observed solar time (known as UT1 and which varies due to irregularities and long-term slowdown in the Earth's rotation). The UTC time standard, which is widely used for international timekeeping and as the reference for civil time in most countries, normally uses precise "atomic time" and consequently would run ahead of observed solar time, unless reset occasionally. The UTC specification requires that this difference should not become significant: the leap second facility exists to provide this adjustment.

When it is mandated, a positive leap second is inserted between second 23:59:59 of a chosen UTC calendar date and second 00:00:00 of the following date. The definition of UTC states that the last day of December and June are preferred, with the last day of March or September as second preference, and the last day of any other month as third preference. All leap seconds (as of 2019) have been scheduled for either June 30 or December 31. The extra second is displayed on UTC clocks as 23:59:60. On clocks that display local time tied to UTC, the leap second may be inserted at the end of some other hour (or half-hour or quarter-hour), depending on the local time zone. A negative leap second would suppress second 23:59:59 of the last day of a chosen month, so that second 23:59:58 of that date would be followed immediately by second 00:00:00 of the following date. Since the introduction of leap seconds, the mean solar day has outpaced atomic time only for very brief periods, and has not triggered a negative leap second.

Because the Earth's rotation speed varies in response to climatic and geological events, UTC leap seconds are irregularly spaced and unpredictable. Insertion of each UTC leap second is usually decided about six months in advance by the International Earth Rotation and Reference Systems Service (IERS), when needed to ensure that the difference between the UTC and UT1 readings will never exceed 0.9 seconds.

This practice has proved disruptive in the modern era, particularly in services that depend on rigorous time stamping or time-critical process control. The relevant international standards body has been debating whether or not to continue the practice with an increasing number of nations supporting its abolition.

Any objections? --John Maynard Friedman (talk) 21:57, 1 November 2019 (UTC)

References

I suggest removing “normally” from the first paragraph: UTC uses SI seconds, by its definition. I also think the last paragraph could be eliminated from the lead, and moved to the beginning ofthe section on problems with leap seconds. If the lead is too long, even with the last paragraph eliminated, I think keeping just the first paragraph would be good: it clearly and succintly describes leap seconds. John Sauter (talk) 07:19, 2 November 2019 (UTC)

Thanks John Maynard Friedman. This looks like an improvement. I support replacing the current lead.

I think we should keep the last paragraph in your proposal as that topic accounts for more than half the body of the article at present. Maybe that's WP:UNDUE or WP:RECENT but that's where we are and the lead should summarize the article. ~Kvng (talk) 16:40, 2 November 2019 (UTC)

Thank you both. I really had doubts about keeping the second paragraph but thought perhaps I might be going too far. I am encouraged to be bold and chop it. Since moves to eliminate the leap second (and the problems they have caused) take up a good chunk of the article, a mention in the lead seems essential to me too, especially as it is only two lines. But "modern era" is a bit pompous, I'll change it to "the twenty first century" (I would prefer "recent years" but that would just attract a "how recent" tag!)--John Maynard Friedman (talk) 21:52, 2 November 2019 (UTC)

All my proposed changes complete

I have replaced the lead as discussed above, with a few minor improvements in the process. I moved the second paragraph of the draft (about the process) down into a new sub-section of that name in the body. The rest of the original lead just duplicated material that was already in the body and better written, so I've discarded it. With great reluctance and fighting my Dr Strangelove impulses, I resisted the temptation to change the title of the History section to A brief history of time. The article is now thrown to the wolves, I rest my case. --John Maynard Friedman (talk) 22:43, 2 November 2019 (UTC)

Source does not support statement in "Slowing rotation of the Earth" section

The last sentence of the second paragraph in the section states, "Over the last few centuries, the periodic component reduced the rate of lengthening of the mean solar day to about 1.4 ms per century," citing a source by Steve Allen, which contains the statement, "... the LOD [length of day] change is 1.4 ms/day/century, which accumulates to 25.6 s/century² ...". The figure of 25.6 seconds per century seems to be in line with the number of leap seconds added in the last few decades. The superscript "2" implies "century squared", but I don't understand what that means. Am I interpreting the source correctly to think that the sentence in the article should read, "Over the last few centuries, the periodic component reduced the rate of lengthening of the mean solar day to about 25.6 seconds per century"? - Donald Albury 18:48, 30 January 2020 (UTC)

Yes, that makes a lot more sense. 1.4ms/century would mean a leap second every 700 years or so! (Could the superscript 2 be a footnote or endnote marker?) --John Maynard Friedman (talk) 20:32, 31 January 2020 (UTC)

I wish I were more confident about interpreting the formula, but it has been close to 60 years since my physics courses. - Donald Albury 00:59, 1 February 2020 (UTC)

The superscript 2 is correct because the long term relationship between Terrestrial Time (TT) (a uniform time) and Universal Time (UT) (solar time on an Earth that is slowing down) is a parabola, not a straight line. The relationship is TT − UT = ∆T. F. Richard Stephenson has studied this since at least 1984 where I found the length of day (LOD) described as 1.4 ms/day/century since 1600. Thereafter it is no longer mentioned by Stephenson. LOD must be integrated to obtain ∆T. Multiplying 1.4 ms/day/century by 36525 days/Julian century gives 51.1 seconds/century². The LOD is a coefficient of t centuries, which when integrated is ⁠t²/2⁠, involving 1⁄2 giving 25.6 s/cy². The vertex of the parabola is at about 1825 where the lengthening mean solar day crosses the day of 86,400 SI seconds. This is the midpoint of the many observations (1790–1892) used by Simon Newcomb to determine the length of his mean solar day (later called the ephemeris day) which he used as the basis of his Tables of the Sun. Both of the parabola's branches, future and past, are positive upward. In his latest publication, Stephensen (2016) gives an updated parabola for the average for 720 BC to AD 2015:

∆T = −320.0 + (32.5±0.6)(⁠year−1825/100⁠)² s

The closest he comes to describing an average after 1600 is a 1500-year oscillation in the LOD:

lod(1500) = +1.78t − 4.0sin(2π(⁠t/15⁠)) ms

where t is centuries after 1825. This must be integrated as before, yielding a long period cosine:

∆T(1500) = +32.5t² + 0.004(⁠15/2π⁠)cos(2π(⁠t/15⁠)) s

Getting back to leap seconds, evaluate ∆T at 2020 and 1972 and subtract, yielding 124 − 70 = 54 s, the number of leap seconds that should have been added in that period if the Earth had slowed down at the average historical rate between 720 BC and AD 2015. These figures ignore the vertex of −320, but the result is the same, 54 s. LOD can be integrated between the same limits, 1.95 cy and 1.47 cy, yielding the same results. These results imply that Earth has recently slowed down less than its historical average because the actual number of leap seconds is only half what the historical average predicts. — Joe Kress (talk) 04:42, 1 February 2020 (UTC)

I got that the rate of lenghtening has been changing, but it has also been close to 60 years since I took calculus (which was not my best subject). So, would it be appropriate for the article to say, "The rate at which the length of the day increases varies. Consideration of the interaction between the Earth and the Moon predicts a rate for the lengthening of the day of 42 seconds per centruy. Over the last two millenia, the rate at which the length of the day has increased has been slower, an average of 31 seconds per century, and in the last four centuries, 25.6 seconds per century."? - Donald Albury 12:31, 1 February 2020 (UTC)

No. All figures needs a superscript 2. The first, 42 seconds per century², is the major portion of the lengthening day due to tidal friction of the Moon on Earth. The second figure, 31 seconds per century² is an old figure which must be changed to Stephenson's latest figure of 32.5 seconds per century². This figure includes both tidal friction (slowing down) and glacial uplift (speeding up). The last figure, 25.6 seconds per century², is also old and must not be used. Changing the figures now in the article to 32.5 seconds per century² is challenging. All figures need a superscript 2 because they are all coefficients of the parabola's t², so its dimension of "century²" must be cancelled by a "per century²" in its coefficient. — Joe Kress (talk) 17:20, 1 February 2020 (UTC)

I found Stephenson's latest cubic splines in the electronic supplements to his main paper cited above. They are 55 smooth cubic polynomials curve-fitted to his data, forming an undulating ∆T parabola, plus a minor conversion for the equivalent undulating straight-line length-of-day. Each polynomial covers a long period where data is sparse, and they cover a series of short periods (3–5 years each) during the modern period of 1800–2016 where data is abundant. This makes the modern curve quite noisy on century time scales and useless for our purposes. I suspect we may have to include Stephenson's old post-1600 25.6 seconds per century² estimate alongside his historical (−720 to 2016) 32.5 seconds per century², as well as the tidal-friction 42 seconds per century² for explanation. I also don't trust my integration of the 1500-year sinusoid so it is out for now. — Joe Kress (talk) 20:44, 2 February 2020 (UTC)

No, 25.6 s/century^2 isn't the change in the length of day, it's the coefficient of the quadratic term in the approximation of ∆T (Delta T), which is a different quantity. The sentence as is indicates that over the past few centuries, the day has lengthened by an average of 1.4 ms each century, which is essentially correct and matches with the source. Remember also that a change in the length of day of 1.4 ms would mean that each century would be longer by 36525 * 0.014 or about 51.2 seconds (half of this gives the coefficient described above for the integration). Arcorann (talk) 11:32, 5 February 2020 (UTC)

Negative vs Positive Leap Second

Ladies, Gentlemen,

Please let me tell you that according to IERS/EOC plot, from early June since May 2020, UT1-UTC increases. Under this condition negative leap second at the end of 2023 the decade should be expected. Anyway, today, negative leap second is more probale than positive one. Then I propose some reference in the article.

With regards and friendship, Georges Theodosiou, The Straw Man Georges T. (talk) 14:04, 16 September 2021 (UTC) 14:20, 18 September 2021 (UTC) 13:53, 21 September 2021 (UTC)

Right now, that interpretation (though reasonable) is wp:Original research for as long as it is you who is drawing that inference. We have to wait until a seriously reliable source says so: a scientific journal not a newspaper. My own guess is that there will be very strong political pressure to avoid doing it, because of the chaos it will cause in a wide variety of systems because it will mean that events that are actually a second apart will be recorded as having occurred simultaneously. --John Maynard Friedman (talk) 14:59, 16 September 2021 (UTC)

Mr. John Maynard Friedman, please let me express you many thanks for you commented my proposal. Also, please feel free to delete this section. With regards and friendship, Georges Theodosiou, The Straw Man, Georges T. (talk) 13:21, 17 September 2021 (UTC)

Mr. John Maynard Friedman, please let me tell you that now, after my edits in title and in my first message, it is Mr. John Sauter's Original Research. Just my latest research agrees with his. Georges T. (talk) 12:21, 18 September 2021 (UTC)

I think a more likely date for the negative leap second is around the end of this decade. I also predict that there will be no serious effort to abolish leap seconds, in spite of the chaos that will be caused by poorly-written software, until three or four months before it happens. My evidence is the similar problem with the year 2000: in spite of the fact that everyone saw it coming, some were unprepared, and there were some glitches due to software that couldn't handle it. My hope is that there are some people who remember the year 2000 and will devote some effort between now and then to testing and fixing the important software. John Sauter (talk) 14:19, 17 September 2021 (UTC)

It is not just an issue of software: for seriously time-stamp critical applications, there is no obvious workaround. Well not that I can see anyway but greater minds that mine may find a way. So my guess is that nothing will be done: world-wide consensus will be wait and see if it rights itself over a 25 year horizon and just ignore these short term perturbations. In almost all contexts, atomic time is preferable anyway: until the deviation exceeds say five minutes, hardly anyone will notice the difference from solar time. Maybe there will be a "UTx" for solar time but it won't be adopted as civil time anywhere. --John Maynard Friedman (talk) 15:06, 17 September 2021 (UTC)

Serious time-stamp critical applications should be written to work in the presence of positive and negative leap seconds. In today's world, everything is software, and the software can be written correctly. It isn't actually hard to do, it just takes the will to do it. John Sauter (talk) 02:49, 18 September 2021 (UTC)

Mr. John Sauter, please let me tell you that since negative leap second has never been introduced, it is terra incognita for the engineers, as in programming "undefined behavior" and "unexpected results" should always be expected. That's the problem. Georges T. (talk) 12:21, 18 September 2021 (UTC) 13:53, 21 September 2021 (UTC)

According to "Future of Leap Seconds" website, there are potential legal issues that would occur in countries like USA, Great Britain, and Canada, that still regard GMT (i.e. UT1) as legal time. For example, an accident recorded on camera, occurring seconds from midnight, may or may not be subject to insurance payout if insurance expires on that midnight. So I would say there are definitely potential issues, and they are not necessarily related to software only. Morycm (talk) 00:22, 1 March 2024 (UTC)

Mr. John Sauter, please let me express you my many thanks for you commented my proposal. Regarding situation in 2000 I post this plot. It is clear that there was not any increase similar to that since June 2021. By the way, let me suggest you and every interested this IERS/EOC page for retrieving data and plots. My way for predictions under aforementioned condition is that in 2 months (June 17 to August 17) increase was 44 msec that is 22 per month. With regards and friendship, Georges Theodosiou, The Straw Man Georges T. (talk) 14:52, 17 September 2021 (UTC)

The Earth generally speeds up during Northern Hemisphere Summer but slows down in Winter, so projecting using just Summer data will give you too early a date for the negative leap second. My reference to the year 2000 was to software and hardware which assumed that the first two digits of the year would always be 19. John Sauter (talk) 02:43, 18 September 2021 (UTC)

(As in the year 2000 problem.) Guy Harris (talk) 08:53, 18 September 2021 (UTC)

Mr. John Sauter, please let me agree on that "The Earth generally speeds up during Northern Hemisphere Summer but slows down in Winter, so projecting using just Summer data will give you too early a date for the negative leap second". Whoever as it seen in this plot, UT1-UTC bottomed on May 2020. Since then it rises in mean terms. Its rising ratio is about 70 msec per year. Under this condition you are right on that "a more likely date for the negative leap second is around the end of this decade", my estimation is the end of 2033 2032, indeed if ITU fail solve the problem till then. With regards and friendship, Georges Theodosiou, The Straw Man, Georges T. (talk) 09:28, 18 September 2021 (UTC) 07:51, 27 September 2021 (UTC) 07:43, 8 October 2021 (UTC)

@Georges T.: For future reference, please do not edit your comments on a talk page the way you did above after other users have replied to those comments. Any edits you make to your comments after discussion has started should be accomplished by striking out (using <del>...</del>) the words you want to replace and inserting (using <ins>...</ins>) the words you want to substitute in. See WP:TALK#REVISE for details. - Donald Albury 13:40, 18 September 2021 (UTC)

Mr. Donald Albury, please let me express you my sincere gratitude for your suggestion. With regards and friendship, Georges Theodosiou, The Straw Man, Georges T. (talk) 14:26, 18 September 2021 (UTC)

Ladies, Gentlemen, please let me report you (IERS snob my messages) something strange. Despite their name (International EARTH ROTATION Service), they do not (actually they do as Δω₃, I apologize) publish in their proper page data and plots for EARTH ROTATION speed (omega). Though omega is UT1's independent variable according to Newcomb's formula (omega replaces sidereal time), UT1-UTC depends on two speeds: omega and cesium atom's oscillation. It follows, monitoring UT1-UTC (or UT1-TAI) we do not monitor omega. With regards and friendship, Georges Theodosiou, The Straw Man Georges T. (talk) 07:25, 29 September 2021 (UTC) 09:50, 27 April 2022 (UTC)

Keeping an eye on Earth's Omega

Ladies Gentlemen, please let me tell you that in IERS/EOC data plots (Georges T. (talk) 10:49, 2 February 2023 (UTC)), Δomega and ΔLOD (from this page), also from IERS/RS-PC (last plot) Georges T. (talk) 08:28, 28 February 2023 (UTC), supported by Vondrak filter's output (indeed by "LOD / 86400 s SI" they mean "LOD - 86400 s SI". They practice "je m'en fous" Georges T. (talk) 13:58, 27 February 2023 (UTC) Georges T. (talk) 13:34, 24 March 2023 (UTC) Georges T. (talk) 07:35, 27 March 2023 (UTC)) one can see that in late 2022 earth is decelerating making positive leap second more probable than negative. Indeed only God knows how long it will last. With regards and friendship Georges Theodosiou. Georges T. (talk) 16:48, 15 December 2022 (UTC)

The Earth's rotation rate is seasonal. It is clear from the charts you referenced that the Earth spins faster during Northern Hemisphere summer.s. To get a useful prediction of the future rate of rotation of the Earth you need to look at a year's worth of data.

The IERS has predicted that the Earth's rate of rotation will continue to be slower than one per 86,400 seconds over the next year. My guess, based on projecting the IERS prediction further into the future than one year, is that we will have a negative leap second somewhere around the end of this decade. John Sauter (talk) 15:02, 17 December 2022 (UTC)

Dear Sauter, please accept my thanks for you answered my message. In Vondrak filter's output it's clear that earth is decelerating in second semester 2022. It is supported by Δomega and ΔLOD graphs. In Δomega, zigzag in October-November is clear lower than in same months last year. In ΔLOD is clear higher. Both mean deceleration independent of tidal and seasonal variations, a random, unexpected, deceleration. With regards and friendship, G. Theodosiou The Straw Man. Georges T. (talk) 10:07, 20 December 2022 (UTC)

Although this trend is clear to you, it is not clear to me, or to the IERS, which continues to predict that over the next year, at least, the length of the day will continue to be less than 86,400 seconds. If you have a mathematical basis for your prediction that the length of the day will increase to over 86,400 seconds enough to cause a positive leap second, perhaps you could present it here. We can then compare your predictions of the length of day with the measured values reported to the IERS. If your predictions turn out to be more accurate than the IERS predictions you will gain credibility. John Sauter (talk) 16:34, 20 December 2022 (UTC)

Mr. Sauter, please let me tell you that credibility's worth is 0 and < 0. With regards and friendship, G. Theodosiou, The Straw Man, Georges T. (talk) 07:08, 28 March 2023 (UTC)

Mr. Sauter, please let me tell you that I predict all of the remaining leap seconds before 2037 will be positive, based on the periodic metonic fluctuation of ΔΩ. With regards and friendship, G. Theodosiou, The Straw Man Georges T. (talk) 12:10 5 April 2024 (UTC), Georges T. (talk) 13:45, 8 April 2024 (UTC).

Thank you for your definite and clear prediction. It will be interesting to see if you are correct. John Sauter (talk) 16:09, 7 April 2024 (UTC)

Ladies, Gentlemen, please let me report you an important error in IERS's UT1-UTC data averaged for tidal variations ("Remove tidal variations" checked), in respect to original. Compare ΔLOD data, original and averaged. Latter are mean values of the former. Georges T. (talk) 12:34, 16 August 2023 (UTC). With regards and friendship, G. Theodosiou, The Straw Man Georges T. (talk) 07:49, 2 August 2023 (UTC)

Ladies, Gentlemen, please let me tell you interesting facts regarding IERS's monitoring of Earth's omega. At IERS/EOC's front page the plot depicts LOD as having equal value in the middle of the year and a year earlier. In the IERS/EOC and IERS/RS-PC (last plot) charts, the zigzags in the middle of the 2023 and 2022 should be at the same height if it were true, but they are not. On the other hand, if you click on their Vondrak filter chart, a back page appears. It is slightly different from the front one and might be accurate. The curve at midyear is still higher than the curve at midyear 2022. By the way, I should mention that I used a paraphrasing tool this time. Please offer your opinion. With regards and friendship, Georges Theodosiou, The Straw Man, Georges T. (talk) 14:30, 1 September 2023 (UTC). Now, the front and back charts on the IERS/EOC website are identical. Those are IERS/EOC. Georges T. (talk) 14:39, 1 September 2023 (UTC)

Ladies, Gentlemen, please permit me to point out that there are clearly three metonic cycles (1964–1984, 1984–2004, 2004–2024 ) in this and this plots. The fact that there is continual acceleration is clearly evident. With regards and friendship, Georges Theodosiou, the Straw Man, Georges T. (talk) 13:49, 4 April 2024 (UTC)

The continual acceleration is not evident to me. It seems more like randomness. Perhaps you can explain so I can understand. John Sauter (talk) 11:54, 7 April 2024 (UTC)

Mr. Sauter, please let me express my thanks for you replied my message and more for you stated your view. I agree, it is a long-term random acceleration. With regards and friendship, G. Theodosiou, The Straw Man, Georges T. (talk) 13:51, 8 April 2024 (UTC)

Let me explain my prognosis, Ladies and Gentlemen, for leap seconds up to 2037. I notice short-term accelerations and decelerations in the earth's Δomega signal, irregularities according to Dr. Bizourd IERS/EOC Directior, randomness according to Dr. Sauter, and noise according to DSP jargon. They are unable to undermine my forecast. In fact, it will be discouraged if a new long-term acceleration appears. With regards and friendship, G. Theodosiou, The Straw Man, Georges T. (talk) 09:07, 11 May 2024 (UTC)

A useful page of links to the future of leap seconds

• Allen, Steve. "UTC might be redefined without Leap Seconds". University of California Observatories, Lick.

The interesting question is whether the question will be left to the astronomers to decide. IMHO (as already noted above), the prospect negative leap seconds (where an event at 23:59:59.9 is followed 200ms later by an event at 23:59:59.1) will be met by a "chorus of disapproval", to put it mildly. --16:58, 19 September 2021 (UTC)

That isn't the way a leap second works. Rather, 23:59:58.9 is followed 200 milliseconds later by 00:00:00.1 of the next day. John Sauter (talk) 05:00, 20 September 2021 (UTC)

Whoops of course it is. I really did not have my brain in gear when I wrote that. Scrubbed. --John Maynard Friedman (talk) 08:15, 20 September 2021 (UTC)

BTW, 23:59:59 UT is 15:59:59 in California and 07:59:59 in Beijing. So maybe nobody would be too bothered in London at midnight New Year's Eve but others certainly would. --John Maynard Friedman (talk) 19:17, 19 September 2021 (UTC)

and of those, this one is the most relevant to the discussion above – though it makes no mention of negative leap seconds – is this one:

• Allen, Steve. "Issues involved in computer time stamps and leap seconds". University of California Observatories, Lick.

Food for thought! --John Maynard Friedman (talk) 19:17, 19 September 2021 (UTC)

Eric Kvaalen proposal: WP:COI vio

I deleted this edit by Eric Kvaalen because of WP:COI violation:

Another suggestion that has been made is to redefine the second, or rather, to intorduce a "civil second" slightly longer than the SI second so that there would be 86400 civil seconds in a mean solar day. This civil second would be redefined every few decades as the rotation of the earth continues to slow down.

Of course someone else is entirely free to reinstate it if they consider it suitable for inclusion. (I'm not saying that it is or that it is not, only that Eric can't be the one to add it.) 𝕁𝕄𝔽 (talk) 15:04, 18 November 2022 (UTC)

Notice that I referenced something that was published in New Scientist -- they decided that it was worth publishing, not I. And it's not against the rules to put in a reference to what you have gotten published. I also referenced a letter to New Scientist by someone else. Eric Kvaalen (talk) 10:44, 19 November 2022 (UTC)

It's true that citations to published works may be added by the author who is also a Wikipedia editor (unless the motivation seems to be to sell more copies of the publication, which is not the case here). The "Reliable sources" guideline does not specifically mention letters to the editor as being reliable or unreliable, but comments in the talk page archives generally take an unfavorable view of the reliability of such letters. Jc3s5h (talk) 13:06, 19 November 2022 (UTC)

WP:RSEDITORIAL does say about letters to the editor: [They] are reliable primary sources for statements attributed to that editor or author, but are rarely reliable for statements of fact. Vpab15 (talk) 19:33, 19 November 2022 (UTC)

My apologies, Eric. I should have checked WP:COI first and Jc3s5h is correct. The relevant policy is WP:SELFCITE. I have reinstated your text and your own NS citation, but not the letter. --𝕁𝕄𝔽 (talk) 21:18, 19 November 2022 (UTC)

Turns out that I was right the first time, but for the wrong reason. I failed to spot that both citations were letters. So it has gone again. --𝕁𝕄𝔽 (talk) 23:53, 19 November 2022 (UTC)

References

Of course letters to the editor are not reliable sources for facts. But that's not how I used them. What I said in my edit was that this idea has been suggested, and that fact is proved by the existence of the two letters. I suspect that the idea has been suggested by other people, but I don't have a reference for that. New Scientist is, I think, the most read popular science magazine, so the fact that something has been published by them means that it has had wide exposure. Eric Kvaalen (talk) 04:49, 20 November 2022 (UTC)

I don't think published suggestions are noteworthy enough to include in the article unless the suggestion is published in a more prominent way that a letter to the editor. For example, the CGPM resolutions from 2022 include the statement "encourages the BIPM to work with relevant organizations to identify the need for updates in the different services that disseminate the value of the difference (UT1-UTC) and to ensure the correct understanding and use of the new maximum value." That is the sort of suggestion that would be important enough to include in a Wikipedia article. A way that a suggestion made as a letter to the editor could be worthy of inclusion in an article would be if several scholarly sources commented upon the suggestion. Jc3s5h (talk) 17:55, 20 November 2022 (UTC)

What next?

The General Conference on Weights and Measures on 18 November 2022 decided to abolish the leap second no later than 2035 (reported by Nature). US law recognizes the CGPM as interpreted by the Secretary of Commerce and Secretary of the Navy as the competent authority to define UTC. But the US is also a member nation of the ITU, and the ITU is responsible for defining how UTC is disseminated. So what steps will we be seeing to actually carry out the CGPM decision?

One step that will be needed will be to alter the way WWV and similar time signals disseminate the difference between UT1 and UTC. Currently this is done by lengthening certain time ticks to indicate the number of deciseconds that should be added to or subtracted from UTC to obtain an estimate of UT1. It is only defined for the range ± 0.9 s. Options include changing the format of the signal, eliminating that feature of the broadcast altogether, or shutting down WWV and WWVB altogether, as was proposed during a recent budget fight. Jc3s5h (talk) 13:33, 19 November 2022 (UTC)

At the risk of drifting into WP:NOTFORUM, I suggest that the message "by 2035 at the latest, could be earlier" is telling the technical committees to produce a plan to make it happen. For WP purposes, we can only wait and see how they resolve it. --𝕁𝕄𝔽 (talk) 13:42, 19 November 2022 (UTC)

Trying to avoid WP:NOTFORUM, if we deem the CGPM decision sufficient to assure that the leap second will be abolished, it will change the way we describe leap seconds and UTC in articles. We would have to regard leap seconds as a method of aligning UTC to UT1, used from 1972 until a date to be determined, rather than as a method that will continue indefinitely. And the political reality is that after 2035 UTC might be an offset version of TAI, or might be a variant of Universal Time such that |UTC - UT1| < n, where the value of n has not been decided upon. Jc3s5h (talk) 13:56, 19 November 2022 (UTC)

Yet another possibility is that it will not be possible to get all the details worked out by 2035, so leap seconds will continue until they are. John Sauter (talk) 15:39, 19 November 2022 (UTC)

It seems that Nature is not convinced that the decision is final either. --𝕁𝕄𝔽 (talk) 11:14, 21 November 2022 (UTC)

Slower by a negative amount?

In the History section, shortly after "rubber second": The phrase "slower than the rate of atomic time by −150 parts per 10 (raised to the tenth)" is confusing. Is that dash a "minus sign/negative sign"?

If so: Doesn't slower by a negative amount mean faster? If "slower" is kept, I suggest removing the negative signs from the various phrases in this section. David10244 (talk) 05:32, 24 January 2023 (UTC)

I have changed "slowed" to "offset", using words from the source, to make the meaning clearer. John Sauter (talk) 06:25, 24 January 2023 (UTC)

That sounds better, thanks. David10244 (talk) 05:05, 25 January 2023 (UTC)

Length of day

In the Process section, in the phrase "in 1972 the average length of day was approximately 86400.003 seconds and in 2016 it was approximately 86400.001 seconds, indicating an overall increase in Earth's rotation rate over that time period".

It seems like there's a decrease, not an increase, in the rotation rate, since there's a decrease in the "length of day"... if the length of a (rotational) day is measured in "atomic seconds". But I could be reading this wrong -- not sure. David10244 (talk) 05:44, 24 January 2023 (UTC)

I believe you are reading this wrong. Consider a simpler example: a ball spins at a rate of one rotation every 2 seconds. In order to make the ball spin at a rate of one rotation every second (decreasing the rotation time) you must increase the rotation rate. John Sauter (talk) 06:29, 24 January 2023 (UTC)

Ah, yes, that makes sense. I was backwards on this one. David10244 (talk) 05:06, 25 January 2023 (UTC)

Article date format

I notice the date format in the article is inconsistent, especially in the references. I found the first edit that introduced dates that contained both the day and the month. It used the dmy format so I have put the {{use dmy dates}} near the top of the article. I will look for inconsistent dates and fix them. Jc3s5h (talk) 16:11, 13 July 2023 (UTC)