Talk:Leap second

Broadcasts?

What does this have to do with broadcasts? That is not explained at all. I suppose broadcasters are affected by this, but the intro suggests that they're the main reason and that sounds a bit odd. Isn't the main thing that there is a variation between atomic clock time, which is constant, and astronomical time, which varies because of variations in Earth's rotation? And shouldn't that then be in the intro? At the very least the broadcast thing should be explained better, because I don't get it. DirkvdM (talk) 19:03, 8 January 2009 (UTC)

I've tried to clarify the article. "Broadcast" is meant to describe radio stations that exist only to provide time and frequency information; it is not meant to apply to ordinary radio stations that provide news and entertainment (although those stations may depend on UTC in less obvious ways). --Gerry Ashton (talk) 19:48, 8 January 2009 (UTC)

Right. Melting polar ice should increase the moment of inertia, as the water redistributes throughout the oceans. You'd expect that to slightly slow the Earth. But we're talking about fractions of millisecond per century here. DonPMitchell (talk) 17:08, 7 August 2009 (UTC)

Observing the Sun

I agree that extra gallactic radio sources can be and are used for measuring the Earths rotation, but the Solar day can only ultimately be measured by observing the sun. By way of explanation, if the radio sources ever moved out of syncronisation with the rotation of the earth around the sun, ( an extremely unlikely but not impossible event) then we would have to rely on our observations of the Sun to give us our Solar day. The point I am trying to make is that apparant Solar time can be measured by nothing more complicated than a sundial. Canol (talk) 02:16, 21 January 2009 (UTC)

It is mathematically impossible for an extra-galactic source to be out of synchronization with the Sun. Geometry requires that the number of sidereal days in any year (tropical or sidereal) be exactly one more than the number of solar days in that year (because a year is one revolution of the Sun) regardless of Earth's rate of rotation. This mathematical relationship is now used to convert an observed extra-galactic source into an 'observed' Sun. Furthermore, the extra-galactic source can be observed about a million times more precisely than the Sun can be observed. Thus, seconds are smaller than the precision with which the Sun can be observed in any one year, so the Sun can never be used to determine when leap seconds should be added on a timely basis. Indeed, the last time the Sun was observed to precisely determine time was about a hundred years ago. — Joe Kress (talk) 21:03, 21 January 2009 (UTC)

There are three important terms: appparent solar time (observed with sun dial), mean solar time, and sidereal time (best determined by observing extra-gallactic radio sources). Both apparaent solar time and mean solar time are useful in daily life; apparent solar time is most useful for activities directly involving the Sun, such pointing moveable solar panels. Mean solar time is most useful for activities conducted according to clock time. Sidereal time is most useful when observing any celestial body other than the Sun. --Gerry Ashton (talk) 21:19, 21 January 2009 (UTC)

Hi Joe and Gerry

I'm not explaining myself very well here. Ignoring the maths for a moment, A solar day is based on the rotation of the earth relative to the sun, hence "Solar". Anything used to measure this that does not have a direct reference to the sun, has to be a secondary measurement, allbeit that it could be a very accurate one

One thing that bothers me though that needs inclusion somewhere in this article, perhaps you or someone could comment. I read somewhere, but can't now find the reference, that an experiment was conducted by taking one atomic clock to the North Pole and one to the Equator. After some time the clocks were brought together again and 'hey presto' they read different times due to the Equator clock moving faster relative to the Pole clock. This was due to Einsteins Theory of relativity as the periphery of the earth moves faster relative to the pole. How often would we need a leap second due to the Equator clock running slow? Canol (talk) 21:44, 21 January 2009 (UTC)

I think we could say that all forms of solar time are motivated by the affect of sunlight on daily life, and that over the period of 25 to 100 years, observations of the sun are required to insure that our conversion of sidereal time to mean solar time is sufficiently accurate. On a day-to-day basis, observation of the sun is a poor way to measure time. I understand that Internatial Atomic Time does indeed include relativistic efects, but I don't know the details of the calculations. --Gerry Ashton (talk) 00:11, 22 January 2009 (UTC)

We cannot state that the Sun is observed to determine time without some citation. I am not aware of any such program. All transit circles, which were used to observe the Sun during the 19th century, have now been retired or are in museums. Only telescopes designed to observe the Sun can be used because normal telescopes will be damaged by the intense heat of concentrated sunlight. Mean solar time is a ficticious time that cannot be observed directly, so an entire year of observations of the Sun's true time are needed to obtain a single crude average over a cycle of the equation of time. Even then, the result is about a million times worse than converting sidereal time. About AD 140 Ptolemy stated (Ptolemey's Almagest, tr. G. J. Toomer, III.9) that mean solar time was 360 time-degrees (a sidereal day) plus approximately 0;59 time-degrees (59/21600 sidereal day). The modern relationship between sidereal time and mean solar time (The new definition of Universal Time) has about ten more significant digits.

Canol, you are probably refering to the Hafele–Keating experiment (1971) which flew four atomic clocks on regularly scheduled commercial aircraft around the world once east and once west. They did not describe the aircraft routes, except to thank Pan Am, TWA, and AA. Although latitude was used to predict the gain or loss of the travelling clocks, they explicitly stated that all clocks fixed to Earth's surface at average sea level, regardless of their latitude, keep the same time because the effect of the varying rotational velocity (the Sagnac effect) at different latitudes is exactly canceled by a corresponding difference in surface potential due to the oblate Earth, at least to the first order. However, all are proper times, meaning they only represent TAI at their locations on the geoid. A form of the Sagnac effect is applicable when comparing one such time with the time of another atomic clock elsewhere on the geoid ("News from the BIPM", Metrologia 17 (1981) 69-74). This experiment confirmed that atomic clocks run faster at high altitudes and run slower moving east (with Earth's rotation).

The time of each fixed atomic clock must be corrected for the reduced potential (gravitational and centrifugal) at its altitude above mean sea level. The gravitational potential only depends on altitude, but the centrifugal potential depends on latitude. All times were corrected for the increased potential on the geoid at the beginning of 1977,[1] decreasing the frequencies of all atomic clocks and lengthening each SI second. This was formalized in 1980 by the statement "TAI is a coordinate time scale defined in a geocentric reference frame with the SI second as realized on the rotating geoid as the scale unit."(in above "News") Now each laboratory must steer the frequency of its atomic clocks to the value they would have if they were located on the rotating geoid (mean sea level). Recently, the required fractional frequency shift for the NIST-F1 atomic clock at Boulder, Colorado (1649 m above the geoid) was determined to be −1798.7×10⁻¹⁶,[2] about 1×10⁻¹³ per kilometer of height, which is several orders of magnitude smaller than the frequency shift used during 1968−1971 to steer TAI into UTC instead of using leap seconds (−300×10⁻⁸). Consequently, gravitational redshift changed the length of the SI second itself—the shift was much too small to warrant a leap second. — Joe Kress (talk) 03:48, 26 January 2009 (UTC)

Joe referenced the Aoki et al. expression for UT1, which was published in 1982. In that work, "the new expression is based on the final values adopted for the position of the FK5 equinox". FK5 (meaning fundamental star catalog 5) is based on, among other things, "an analysis of absolute observations of the Sun, planets, and minor planets". Seidelmann describes an update in 2005, but I have not figured out if the update involves any new solar observations. --Gerry Ashton (talk) 22:29, 26 January 2009 (UTC)

Thanks for the citations. A more fruitful search avenue might be literature associated with the national ephemerides such as JPL Solar System Dynamics and the Explanatory Supplement to the Astronomical Almanac. I remember reading that most of the sources used by JPL were various space missions, most of which they controlled. — Joe Kress (talk) 21:16, 27 January 2009 (UTC)

Thanks for the suggeston about JPL. I'm afraid the Explanatory Supplement is rather out-of-date, but according to Seidelmann's home page he is working on a revision. --Gerry Ashton (talk) 21:45, 27 January 2009 (UTC)

The limited view online version claims to be "completely revised and rewritten" (2005). However, all front matter, including its Abbreviated contents, Contents, List of figures, List of tables, Forward and Preface are identical to the 1992 edition, including all page numbers, so it is just a reprint. The blurb was apparently a reference by the publisher to the old Explanatory Supplement to the Astronomical Ephemeris and American Ephemeris (1961–1974). — Joe Kress (talk) 00:09, 28 January 2009 (UTC)

The Explanatory Supplement stated that many transits of the Sun were observed (declination and time of day) by the USNO's six-inch and nine-inch meridian circles during 1911–1975 with a standard deviation of 1.0" which formed part of the observational data upon which the JPL ephemeris DE118 was based.(pp.290–1, p.301) However, it also stated that because the modern definition of UT1 is a function of its rotation angle in space (sidereal time), "UT1 deviates secularly from solar time; however, the divergence is extremely small."(p.51) Thus despite continued observations of the Sun, UT1 will continue to deviate by a very small amount from mean solar time. The defining equation for UT1 (p.50) is equation 13 in Aoki et al., relative to the precessing equinox, effective 1984 January 1:

GMST1 of 0^hUT1 = 24110.54841^s + 8640184.812866^sT_u + 0.093104^sT_u² − 6.2×10^−6sT_u³

where T_u = Julian UT1 centuries since J2000.0.

The equivalent definition for UT1 given in IERS Conventions (2003) 5.4.4 is the Earth Rotation Angle, relative to the Celestial Ephemeris Origin (a fixed or catalog equinox), effective 2003 January 1:

θ(T_u) = 2π(0.7790572732640 + 1.00273781191135448T_u)

where T_u is the number of UT1 days since J2000.0. Because the CEO does not move, no higher order terms are needed. Because the International Celestial Reference Frame is independent of Solar System dynamics, this definition is independent of observations of the Sun. — Joe Kress (talk) 06:13, 31 January 2009 (UTC)

The definition of UT1 is independent of observations of the Sun for now. The definition of UT1 has been changed a number of times during the 20th and 21st centuries; no doubt it will change again if any deviation between solar observations and the current definition makes the current definition inconvenient. --Gerry Ashton (talk) 13:39, 31 January 2009 (UTC)

The group that wants to replace leap seconds with leap hours, which would displace UTC from UT1 by as much as half an hour, would argue that a much smaller disagreement between UT1 and mean solar time is of no concern. — Joe Kress (talk) 21:43, 31 January 2009 (UTC)

Thanks for taking the time and effort to answer me, I was aware of Hafele–Keating and the Pan Am aircraft experiments. If I have interpreted Joe Correctly, he is saying that there is a small time difference between the equator and the pole, but this is not due to the relative speeds at these locations, rather it is due to the earth not being a perfect sphere and so gravity at the pole is different from that at the equator.
I am in London next week and intend spending some spare time at the Greenwich observatory where I will put these questions to the historical experts there. If I learn, or can cite anything more, I'll report back here. Canol (talk) 23:55, 26 January 2009 (UTC)

No. An atomic clock on the equator tries to 'tick' more slowly because it is moving rapidly, but it is also farther from Earth's center of mass so it tries to 'tick' faster. The two effects exactly cancel each other, so atomic clocks everywhere on the rotating geoid keep the same time regardless of their latitude. But atomic clocks above the rotating geoid, like that in Boulder, 'tick' faster because they are farther from Earth's center of mass than the geoid, but have the same speed as the rotating geoid so there is no cancelling slowdown. — Joe Kress (talk) 08:46, 27 January 2009 (UTC)

democratic process regarding proposals to abolish leap seconds

I have been advised to ask this question here rather than in the main article: Speaking as someone who thinks this is a bad proposal and wishes to lobby my appropriate representative if needed, how do I find out who represents me in the appropriate forum and what their position is? —Preceding unsigned comment added by 75.87.135.149 (talk) 05:44, 26 July 2010 (UTC)

You should check this with ITU Working Party 7A (WP 7A) - Time signals and frequency standard emissions. As far as I can see from the 2009 meeting the situation is still inconclusive. --Muhandes (talk) 07:19, 26 July 2010 (UTC)

Leap Second Abolishment Progress Since 2008?

The section "Proposal to abolish leap seconds" appears to be written in 2008. As far as I can tell, there has been activity since then. However, as an outsider to the ITU I am not able to piece together the information well enough to update the page.

This 2011 ITU document is a second questionnaire on the issue. It says, in part:

From the responses to the Questionnaire sent by the Director of the Bureau it appeared that 8 Administrations were in favour of the revision of Recommendation ITU-R TF.460-6 while 3 were against it. Since no consensus could be reached at the SG 7 meeting in October 2010, SG 7 decided to send the Recommendation to the Radiocommunications Assembly, as well as sending a new circular letter reminding Administrations to respond to the questionnaire on this issue.

It requests responses before 19 September 2011, and refers to a 2012 Radiocommunication Assembly as the body that will receive the results of the questionnaire.

It's worth noting that 8 out of 11 (the apparent outcome of the earlier questionnaire) is 72%, exceeding the 70% threshold for abolition given in the current article text. --JeffEpler (talk) 20:46, 20 June 2011 (UTC)

Additional resources on abolishment timeline: Future of Leap Seconds - Recent Events Comission 32 Time --JeffEpler (talk) —Preceding undated comment added 01:28, 21 June 2011 (UTC).

Another graph showing leap seconds

The graph showing leap seconds as the difference between UT1 and UTC is fine, but it doesn't seem like it would be intuitive to the layperson. I have created a graph showing the number of seconds that have accumulated between the UTC, UT1 and the TAI atomic clock standard, established in 1959. One can clearly see the discontinuous nature of UTC through the addition of leap seconds, as well as the fact that it represents an approximation of UT1. It is also interesting to note that the international agencies had to issue corrections several times during a year prior to 1972, before deciding to use leap seconds. The data comes from the http://maia.usno.navy.mil/ser7/ website. On the other hand, perhaps this is one graph too many.

Jdlawlis (talk) 13:36, 16 August 2011 (UTC)

The IERS provides a comparable figure from 1972 to 2009 [3]. Note that their figure shows a negative accumulation, that is, UTC−TAI and UT1−TAI, not TAI−UTC and TAI−UT1 (leap seconds and a slowing Earth cause specific UTC/UT1 times to be less than the equivalent TAI time). — Joe Kress (talk) 05:22, 17 August 2011 (UTC)

Thanks for drawing that webpage to my attention, Joe. Jdlawlis (talk) 19:17, 18 August 2011 (UTC)

Clarification needed

At the bottom of the "Proposal to abolish leap seconds" section the article says:

• 2012: If 70% of member states agree, the Radiocommunication Assembly will approve the recommendation
• 2017: If 70% of member states have voted to abolish the leap second, application of leap seconds will stop and UTC will become a continuous time scale, otherwise they will stay the same.

Given that only 16 out of 192 member states responded, does this really mean "70% of member states" or does it mean "70% of member states who vote/agree"? -- Q Chris (talk) 12:00, 4 November 2011 (UTC)