Talk:Ephemeris time

Untitled
Should the correct title of this article be "Ephemeris time" or "Ephemeris Time"? The article is at the former title, but the text uses "Time". If someone knows, I can change or move it accordingly. -- Infrogmation 14:39, 24 August 2005 (UTC)
 * I believe it should be "Ephemeris time". There were articles under both names, and I merged them together here, so there may be a relic 'Time' left. Salsb 15:31, August 24, 2005 (UTC)

When adopted?
Using the information that ET and TAI were already 32.184 seconds divergent when TAI was adopted in 1958, and a rough guesstimate of average increasing divergence of 0.7 seconds per year, it looks as if the divergence between UT and ET was first recognized around 1912. Could we add information to this article about when ET was first adopted, and when its divergence from UT was first recognized? arkuat (talk) 15:32, 18 May 2008 (UTC)

WP:RD/Science gave me the hint that Simon Newcomb was probably involved. arkuat (talk) 02:59, 19 May 2008 (UTC)

Further info, copied from the Reference Desk:


 * Ephemeris Time (ET) was so named because it was the independent time scale used in government ephemerides from 1900 through 1983, in the American Ephemeris and Nautical Almanac, the British Nautical Almanac and Astronomical Ephemeris, the French Connaissance des Temps, the German Astronomisches Jahrbuch, and the Spanish almanac. ET was defined (but not named) by Simon Newcomb in 1895/98 as the weighted average of mean solar time referred to the Royal Observatory, Greenwich between 1750 and 1892. It was implemented at Greenwich mean noon on 31 December 1899 (0 January 1900), meaning that from that date on the stated positions of the planets in the ephemerides would be calculated in terms of ET, not in terms of Greenwich mean solar time (named UT in 1928) which had been used before 1900. ET differs from UT quadratically, not linearally, that is, the difference is generally a parabola. See ΔT. However, this divergence was not recognized in 1895, instead its implemention included a "great empirical term", a sinusoid with a period of 257 years. — Joe Kress (talk) 08:09, 19 May 2008 (UTC)

Notes about revision of article
Much additional information (citation-sourced) has been added, and the article rearranged. The 'Reference Desk' info about start date 1900 and definition by Newcomb appears to be incompatible with the sources, but owing to inexperience I'm not quite sure where the 'Reference Desk' is or what can be done about it. (Newcomb published in terms of GMT, as was conventional for his time, and additional sources can be cited if required to show that he regarded the evidence for the rotational fluctuations as not yet conclusive.)

Terry0051 (talk) 01:41, 28 February 2009 (UTC)

definition of 'earth's orbit around the sun'
Is 'earth's orbit around the sun' defined in relation to the 'fixed stars' (aka sidereal year - which corresponds to earth's actual position in space relative to the sun and 'fixed stars'), or to the 'tropical year' (which is the modern standard definition of 'year' but corresponds to the length of time between vernal equinoxes, and is slightly shorter than the period of a full orbit around the sun due to the direction of earth's axial tilt rotating counterclockwise over a 26,580 year period? this should be clarified in the article Firejuggler86 (talk) 00:39, 29 October 2020 (UTC)
 * This is explained on page 82 of the Explanatory Supplement to the Astronomical Almanac 3rd ed. It's based on an abstraction of the tropical year. The mean motion of the Sun was calculated by Newcomb at the end of the 19th century. If one takes the derivative of the motion, that is, δL/δT at January 0, 1900, 12 h UT, and calculate how long long it would take the mean longitude of the Sun as viewed from the Earth to increase 360&deg;, the answer is {{val}31556925.9747}} seconds, which implicitly defines the second. The modern definition of the second, in terms of a particular kind of atomic clock, was set up to be as close to the ephemeris second as possible. Jc3s5h (talk) 13:10, 29 October 2020 (UTC)

Ephemeris time based on mean solar day
Yesterday I changed
 * "Although ephemeris time was defined in principle by the orbital motion of the Earth around the Sun"

to
 * "Although ephemeris time was defined in principle by the mean solar day"

with the comment "Ephemeris time was defined based on 36525 mean solar days, not on the actual orbital motion of the Earth around the Sun".

Now has reverted that with the comment "Not true. Read the lead of this article. By the 1950s it was certain that the rotation of the Earth on it's axis was irregular, and the time required for the Earth to orbit the Sun was more nearly constant, so the official definition of time used in the calculation of ephemerides was changed."

Yes, I have read the lead, and I have read much of the article by G.M. Clemence. Yes, I know that it was known that the rotationi of the Earth on its axis was irregular. But the fact is that the ephemeris day was based on the length of 36525 mean solar days. It wasn't based on some figure for the length of a tropical year or of a sidereal year. That's why the number of ephemeris seconds in a day is very close to 86,400 -- it's more or less by definition.

Eric Kvaalen (talk) 05:51, 16 November 2022 (UTC)
 * I will quote some passages from Chapter 3, "Ephemeridies", by McCarthy & Seidelmann
 * "The independent variable in ephemerides is uniform time From the time of the Hellenistic astronomer Ptolemy, the concept of mean solar time was used for this independent variable. Following the recognition of the variability of the Earth's rotation, a new time scale, Ephemeris Time, based on the Earth's orbital motion was introduced. Problems with real-time realization of Ephemeris Time soon became apparent, and atomic time became available as an alternative source of Ephemeris Time. [p. 35]"


 * McCarthy & Seidelmann's Chapter 6 is titled "Ephemeris Time". The development of the definition is described on pages 80 to 83. They describe it as a recommendation at the International Colloquium on the Fundamental Constants of Astronomy in Paris in 1950, followed by a series of improvements until the CIPM put it in it's final form in 1956. When I apply the changes to the starting definition, I come up with
 * In all cases where the mean solar second is unsatisfactory as a unit of time by reason of its variability, the unit adopted should be the tropical year at 1900.0; that the time reckoned in these units be designated Ephemeris Time; that the change of mean solar time to ephemeris time be acomblished by the following correction
 * ΔT = 24.349s + 72.381s T + 2950s T2 =1.82144 B.
 * where T is reckoned in Julian centuries from 1900.0 January 0 Greenwich Mean Noon and B has the meaning given by Spencer Jones in Monthly Notices R.A.S. (Vol 99, 1938, p. 541) and the second is the fraction 1/$31,556,925.975$ of the length of the tropical year for 1900.0
 * A quote from page 86 helps to understand the situation:
 * "Until 1960 the second was defined as 1/$86,400$ of the mean solar day, ignoring the variability in the earth's rotation and assuming that the Earth's rotation was uniform. In 1960 the ephemeris second was introduced as the replacement for the second defined in terms of mean solar time."
 * My interpretation is that the difference between time observed with instruments like those described in the Transit instrument article, and time used in equations to calculate the position of celestial objects, was glossed over. Once ephemeris time was clearly defined, these two different kinds of time could be dealt with more rigorously. Jc3s5h (talk) 14:26, 16 November 2022 (UTC)
 * Hilton & McCarthy explain the lack of a clear modern definition of mean solar time on their page 231
 * "Underlying the concept of mean solar time was the assumption that the rotation of the Earth was uniform. In the first half of the twentieth century, the lunar ephemeris demonstrated that this assumtion was incorrect. Thus the mean solar time was no longer used in precise timekeeping. It was replaced by two somewhat different concepts of time: Ephemeris Time, (ET), was introduced to satisfy the desire for a uniform measure of time, and Universal Time, UT, to measure the earth's rotation. Originally UT was introduced to specify Greenwich mean Time measured from noon instead of midnight. [References to other sections of the book omitted.]"
 * Jc3s5h (talk) 14:50, 16 November 2022 (UTC)


 * James L. Hilton & Dennis McCarthy (2013), "Precession, Nutation, Polar Motion, and Earth Rotation" in Sean E. Urban and P. Kenneth Seidelmann, eds., Explanatory Supplement to the Astronomical Almanac 3rd ed., University Science Books, Mill Valley, CA.
 * Dennis McCarthy & P. Kenneth Seidelmann (2009), TIME From Earth Rotation to Atomic Physics, Wiley-VCH, Weinheim, ISBN 978-3-527-40780-4.

ΔT is NOT parabolic
Quotation from the article: The difference between ET and UT is called ΔT; it changes irregularly, but the long-term trend is parabolic, decreasing from ancient times until the nineteenth century, and increasing since then at a rate corresponding to an increase in the solar day length of 1.7 ms per century.

It's an error made in the past, in the 50's, to assume ΔT is parabolic. If it was, ΔT in 1870 equals that in 1900 and ΔT in 1650 would equal that in 2024. It's obviously wrong.

Furethermore, parabolic ΔT has a minimum point in 1893, which has no physical meaning. 2A01:E0A:9EE:BD20:ED1D:EF5E:5A7B:D74C (talk) 16:58, 30 June 2024 (UTC)


 * The quotation mentions ancient times, but your comment doesn't mention any date earlier than 1650. The cited 2004 paper by Morrison and Stephenson shows a graph on page 330 which contains a best-fit parabola to the available data points and it doesn't look all that bad to me. Jc3s5h (talk) 19:58, 30 June 2024 (UTC)
 * I did not mention an earlier date because 1650 ( 1620 more exactly) has the same ΔT as 2024 according to the parabolic model. Two same ΔT mean that the corresponding UT must be the same, while it is impossible that UT of 1620 equals that of 2024.
 * According to the definition, ΔT = TT - UT, before around 1900, UT went faster than TT and ΔT should be negative, after 1900 UT goes slower than TT and ΔT should be positive. 2A01:E0A:9EE:BD20:DD2A:EDA5:5E4:32DC (talk) 18:30, 1 July 2024 (UTC)
 * A more recent article, by Stephenson, Morrison, and Hohenkerk (2016), provides a fit to observations over the year range -700 to 2000. It is
 * ΔT=−320.0 + (32.5 ± 0.6)((year - 1825)/100)2 (Equation 4.1)
 * I suggest you try plugging your ideas into that parabola and see how it does. I must confess I don't quite understand why the ± symbol is in the equation, unless it is an indication of uncertainty. Jc3s5h (talk) 20:54, 1 July 2024 (UTC)
 * Jc3s5h (talk) 20:54, 1 July 2024 (UTC)

I have read since years almost all of their papers concernaing ΔT. The paper 2016 is the last one but not new. the ± symbol in the equation is an indication of uncertainty. — Preceding unsigned comment added by 2A01:E0A:9EE:BD20:E8B4:7D59:4CEC:B377 (talk) 10:44, 2 July 2024 (UTC)
 * "According to the definition, ΔT = TT - UT, before around 1900, UT went faster than TT and ΔT should be negative, after 1900 UT goes slower than TT and ΔT should be positive." I disagree.
 * Using the table of ΔT in the Astronomical Almanac for the Year 2017 on page K8–K9 ΔT the minimum of ΔT occurs in 1895 (-6.47 s) and the value closest to 0 is in 1902 (-0.02 s). If for discussion we neglect that distinction and pretend the minimum and the 0 value were both in 1900, consider time traveling from 1900. Before then, the length of the UT day was shorter than the TT day. (Treat TT as equivalent to ET for discussion). So a clock set to UT spins faster than a clock set to TT as we move back. In 1641 the difference would have accumulated to 60 seconds. 1600 Jan 1 12:00 UT would be 1 minute earlier than TT, so the same moment would have been 12:01. TT-UT = 1 min = 60 s.
 * Traveling into the future from 1900, the UT day is longer so the TT clock spins faster. The difference accumulates to 60 seconds in 1994. So at 1994 Jan 1 12:00 UT, the TT clock has spun more so reads 12:01. Calculating ΔT = 12:01 - 12:00 = 1 min = 60 s.
 * So ΔT is always positive with the approximations used for the discussion. The table shows in real life ΔT was positive except from 1872 to 1902. Jc3s5h (talk) 13:57, 2 July 2024 (UTC)
 * As you, I was confused at the beginning of trying to resolve this problem. Just one question before going further: what is the ΔT between 1641 and 1994?
 * The two clocks model is quite right. UT clock is a real one, the rotation of the earth. TT (ET) clock is a fictitous one, as like the mean sun, a fictitious sun called by Jean Meeus. 2A01:E0A:9EE:BD20:10F7:17CA:640A:30AD (talk) 15:01, 3 July 2024 (UTC)
 * The values given in the Astronomical Almanac are + 60 s for 1641 and + 59.98 s for 1994. The values in the table are computed differently for different year ranges, depending on what data was available for the ranges.
 * The pages with the data are only available by buying the paper publication, and it would be a pain to try to add those values here. There is another high-quality source online for the years 1846 to now at https://www.iers.org/IERS/EN/DataProducts/EarthOrientationData/eop.html
 * Also, I have seen online images of the Astronomical Almanac, since much of it is not copyrighted. You could search for that. Jc3s5h (talk) 15:11, 3 July 2024 (UTC)
 * Looking more closely, I see the data from IERS is only valid starting in 1956, when atomic clocks came into use. There is an academic collaboration that has put the almanac online at https://babel.hathitrust.org/cgi/pt?id=uc1.31210021820236&seq=7 Jc3s5h (talk) 15:42, 3 July 2024 (UTC)
 * Our discussion does not need a high precision of ΔT. Accoding to the parabolic model, the ΔT between 1641 and 1994 is zero or near zero. It means that UT in these two years is the same. Obviously, it is not true. 2A01:E0A:9EE:BD20:10F7:17CA:640A:30AD (talk) 20:36, 3 July 2024 (UTC)
 * Earlier you asked "Just one question before going further: what is the ΔT between 1641 and 1994?" This is not a valid question. ΔT is the value of TT at a certain moment minus the value of UT at that same moment. The  ΔT btween 1641 and 1994 is meaningless. Jc3s5h (talk) 23:05, 3 July 2024 (UTC)
 * "ΔT = 60 s in 1994" means that comparing to "ΔT = 0 in 1900", "ΔT = 60 s in 1994". One can compare the ΔT of any two time points.
 * We can regard things from a difrerent angle. dΔT/dt is the difference of the two rotation rates of the two clocks. A decreasing dΔT/dt means an increasing dUT/dt, but we know that the UT clock slows down during 1600 to 1900. 88.167.186.244 (talk) 07:14, 4 July 2024 (UTC)
 * Correction: We can regard things from a difrerent angle. dΔT/dt is the difference of the two rotation rates of the two clocks. A decreasing dΔT/dt means that the UT clock run slower than that of TT clock, but we know that the UT clock run faster during 1600 to 1900. 2A01:E0A:9EE:BD20:7CB9:6263:F7DF:6A82 (talk) 08:48, 9 July 2024 (UTC)


 * Stephenson FR, Morrison LV, Hohenkerk CY. 2016 Measurement of the Earth’s rotation: 720 BC to AD 2015. Proc. R. Soc. A 472: 20160404. http://dx.doi.org/10.1098/rspa.2016.0404