Talk:International Atomic Time

Leap second scheduled for the end of the year
There is a leap second scheduled to occur at the end of the year (2005-12-31T23:59:60), so that the article should probably be changed to read UTC = TAI - 33 seconds somewhere around that time.

The announce can be found at http://hpiers.obspm.fr/iers/bul/bulc/bulletinc.dat, or linked via http://tycho.usno.navy.mil.

Real time
Does the use of the term real time have any meaning or usefulness in the context of TAI? —Preceding unsigned comment added by Martin Hogbin (talk • contribs) 13:18, 16 November 2008


 * TAI is not related to a real time clock within a computer, either in the sense of technology (no quartz crystal) or in the sense of a simple multiple of a second, like a power of two. Nor is it related in the sense of civil time because it now differs from that by over half a minute. Nor does "real time" apply in the sense of the present, because TAI is also used as a timestamp for any event, whether it occurred in the past, present or future. "Real time" does not apply as a type of time, more appropriate terms being either a coordinate time on the geoid or a proper time. I can't think of any meaning of "real time" that applies to TAI. — Joe Kress (talk) 20:40, 16 November 2008 (UTC)

Broadcast realisations of TAI are real-time in the sense that they are available immediately: they tell you (to within some uncertainty) what the TAI time is *now*. TAI itself (with highest precision) is not real-time in this sense, because it's determined retrospectively. 195.81.245.98 (talk) 10:03, 22 May 2013 (UTC)

"Realisation of Terrestrial Time"
I don't understand what it means to say that TAI is the "principal realisation of Terrestrial Time". I'm not even sure what "realisation" is supposed to mean here. It seems to imply that TAI is in some way based on or derived from TT, which it is not. It's the other way around. Terrestrial Time is defined as TAI + 32.184 seconds, with the offset inserted solely to provide continuity between Terrestrial Time and a former astronomical standard called Ephemeris Time, which TT replaced due to its increased accuracy. —Preceding unsigned comment added by 91.125.107.176 (talk) 19:59, 21 April 2011 (UTC)


 * I think the answer to your question is that TT is a theoretical ideal, which real clocks can only approximate. TIA (+ 32.184 seconds) is the best representation, based on real clocks, of this theoretical time available. This does not imply that TIA is based on TT, indeed it cannot be since TT is only a theoretical time. Martin Hogbin (talk) 21:41, 21 April 2011 (UTC)

Terrestrial Time is defined as a theoretical ideal; the definition itself contains no reference to TAI. See its article and the IAU resolutions referenced from the first sentence of the article. Particularly, in the 1991 resolutions, note 1 on Recommendation IV: "... As the errors in the realization of TAI are not always negligible, it has been found necessary to define an ideal form of TAI...". Also note 5 discussing divergence between TAI and TT, and note 9 on notation for realisations of TT with particular reference to TT(TAI). Conversely, TAI is based on TT, in the sense that it's a technological attempt to run clocks in a way that matches the ideal TT. This is what is meant by "realisation of Terrestrial Time". TT(TAI) = TAI + 32.184 s is a close approximation of TT, but closer approximations are available. 195.81.245.98 (talk) 10:24, 22 May 2013 (UTC)

The user who posted the comment at the start of this section actually edited out the "principal realisation" phrase that ey didn't understand, and replaced it with an incorrect statement that TAI "is the basis ... for Terrestrial Time". That statement is the polar opposite of the truth: in fact TT is the basis for TAI. Regrettably I didn't notice the incorrect edit back then, six years ago, but yesterday I spotted it and restored the original correct statement ("TAI is the principal realisation of TT"). That edit was since reverted, returning to the incorrect statement, by an editor who suggests that the "realisation" language would require that TAI approximate TT with no offset. Of course it doesn't do that, because of the 32.184 s offset. The offset doesn't really invalidate TAI being a realisation of TT, though this does make it a slightly looser sense of "realisation" than the strictest available (TT(TAI) is the realisation of TT in that stricter sense). TAI is a realisation of TT in the very useful sense that it is a technological attempt to run clocks to tick at exactly the rate prescribed by the ideal TT. I have now restored the "principal realisation" statement, with a qualifying parenthetical acknowledgement of the fixed offset. If this somehow doesn't satisfy everyone, then at least do not just revert my edit and restore the completely incorrect statement that TAI is the basis of TT. Rather than do that, you should either reword the "principal realisation" statement in a more satisfactory way, or failing that remove the statement entirely and leave no mention in the summary of any relationship to TT. 2001:8B0:80D:0:0:0:0:1 (talk) 11:03, 4 May 2017 (UTC)

Added Archives
Here is a list of archives/accessdates that I added to this article. -- Tim 1357  talk 04:43, 25 September 2016 (UTC)

Early timescales
has reverted my attemt to make the discussion in the History section consistent with itself. I don't beleive my edit changed what was being said here so if the contention in Jc3s5h's edit comment is correct, there is still a problem with the discussion of the relationship between UT2 and early time scales. And the discussion is not self consistent. ~Kvng (talk) 22:36, 16 November 2018 (UTC)


 * That section of the article does need some improvement. It names some early atomic time scales (at least, I infer they were atomic because of what those groups were working on, plus the name of a particular atomic clock). But it then just stops talking about the early atomic time scales without explaining how the early time scales evolved into TAI. Also the section does not explain whether those early time scales were discontinued, or if some of them survive, as some sort of working time scale used in the construction of TAI. I'll look at an appropriate book and see if I can improve it. Jc3s5h (talk) 22:49, 16 November 2018 (UTC)
 * I've made changes that I hope provide sources for everything and make the level of detail more even. I was not able to find dates when the time scales from institutions other than BIPM faded away.
 * As for the relationship between early time scales and UT2, several of them, as I clarified in the article, chose to synch their scale with UT2 at Julian Date 2436204.5 (1 January 1958 00:00:00). In reading McCarthy and Seidelmann, I see mention of preserving the continuity of the respective time scale whenever improvements in the scale were made. Potentially, the synchronizations were done based on different sets of astronomical observations of UT2 and different computations, so the three time scales might nominally have the same epoch, but if we were to reconstruct their epochs using the best modern methods, they would probably turn out a bit different. But I don't have a source that would let me put a statement along those lines in the article. Jc3s5h (talk) 00:21, 17 November 2018 (UTC)
 * I did my own calculation using the definition of UT2 and a value of Delta T at January 1, 1958 of 32.166 seconds. This value comes from the US Nautical Observatory's Multiyear Computer Interactive Almanac. I found that 1958 Jan 1 00:00:00.000 TAI = 1958 Jan 1 00:00:00.013 UT2. Jc3s5h (talk) 01:09, 17 November 2018 (UTC)

Whats the time?
Is there a website where I can read the actual TAI? —185.104.138.31 (talk) 15:31, 29 May 2023 (UTC)