Do you develop on GitHub? You can keep using GitHub but automatically sync your GitHub releases to SourceForge quickly and easily with this tool so your projects have a backup location, and get your project in front of SourceForge's nearly 20 million monthly users. It takes less than a minute. Get new users downloading your project releases today!
Posted
by
timothy
from the seconds-all-taste-the-same dept.
ygslash writes "The IERS has announced today that, after seven years, there will once again be a leap second this year. On December 31, 2005, the time 12:59 will last for 61 seconds."
This discussion has been archived.
No new comments can be posted.
The thing that bothers me is, when that big quake in Indonesia went off last year and caused that big tsunami, they talked about how the Earth's rotation SPEEDED UP. If it has already slowed down again in only one year, such that a leap second is needed, then that implies some other place has been bulging and may be about to give way. So, any Slashdotters who can pass this inference on to the geology folks, please do so pronto! Thanks!
I have calculated[1] that in 1000 years a leap second will be required about every two months. It's likely that at that time we would still be using time standards similar to those in use now.
On the other hand, in 1 million years, about 15 leap seconds will be required each day. Therefore, at some point timekeeping must necessarily divide the day into units that are not an integral number of seconds. We would have a situation where the record for the 100 metres dash is expressed in seconds, but the length of the second used for dividing up the day is not the same length. Such "stretched time" has already been used for the Spirit and Opportunity rovers on Mars.
[1] A common formula for approximating the evolution of delta-T over time is 31 * Cy^2, where Cy is expressed in centuries.
Not so many centuries ago, the concept of 'hour' was flexible, depending on the season. 12 day hours, 12 night hours, regardless of the ratio between them. Back to the old?
The same uber-nerds who defined the meter have also defined the second.
Lifted from wikipedia - The second is one of seven SI base units. It is defined as 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 at zero kelvins.
Now that we have a definition perhaps someone could tell us what it means.:)
All it means is that someone looked at the conventional length of a second (1/86400 of a day) and found a natural phenomena that was really, really, really, really, really close, and unlike the length of a day, won't change over time.
As another example, take the metre. You can see how the definition of the metre became more and more precise here [wikipedia.org]. I don't see it mentioned there, but the original "meaningful" definition of the metre, if extremely imprecise, is based on water: 1 ml of water = 1 gm in mass =
That's incorrect. A metre was once defined as one ten-millionth of the distance between the equator and the north pole. The polar circumference was chosen because a polar circumference passes through Paris.
The mass definition was thus based on the length of the metre, not the other way around.
Ah. I always thought that the water relationship was the first "conventional" definition -- in the same way as units like inches and cubits have positively banal original definitions (inch = length of three barley corns laid end to end, or the length of a thumb; cubit = length of the forearm), and that they only switched to a complex definition to ensure precision (and over time, progressively more complex -- and in human terms, meaningless -- definitions in order to increase precision).
The second is one of seven SI base units. It is defined as 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 at zero kelvins.
The definition is made in terms of the most accurate way we have of measuring it - with the atomic clock. There's a description of how they work here [howstuffworks.com].
In the future the definition may change - there are developments to produce "optical clocks" which are more accurate
I expect that as the concept of the "year" becomes more and more antiquated and less and less meaningful, we will begin to simply count some arbitrary number and due without the likening to arbitrary phenomena.
You dont think things will change in 1000 years? It's hard to keep things the same for 10. And that will just get more tedious once we abandon the "year", of course.;)
I have calculated[1] that in 1000 years a leap second will be required about every two months. It's likely that at that time we would still be using time standards similar to those in use now.
Simple solution, change the 60 Hz power line frequency to 59.9999885922 Hz, causing power-line synchronous clocks to slow down to match the daily rotation of the Earth (as referenced to the Sun).
On the other hand, in 1 million years, about 15 leap seconds will be required each day.
They'll just adjust the placement of the leap year, which is much simpler than doing constant leap second adjustments.
Dunno if that's an attempt to get a funny mod or what, but I'll take it seriously.
There's only one slight problem with this, noontime as the clock will happen an hour and a half earlier in the day every year, until after eight years noon occurs in the middle of the night. After another eight years of this silliness, you can add an extra Leap Da
A common formula for approximating the evolution of delta-T over time is 31 * Cy^2, where Cy is expressed in centuries.
The formulae are to some extent empirical as well as being approximations. The evolution of delta-t is also extremely 'noisy', and is far from a good fit with any of the formulae [phys.uu.nl].
For the last few years it's been at around 64 seconds (see data in here [navy.mil]), which comes to about 3 or 4 seconds less for the present time than some recent formulae were predicting even only a few years ago, and l
It's a bit optimistic to extrapolate over 1000 years!
This is why I made sure to state that the formula was an approximation.
Although it is unlikely that such formulae are going to be accurate 1 million years form now, it does not detract from the point that I raise, namely that our current definition of 86400 seconds to the day and 1 second = 9192631770 cesium transitions cannot both last indefinitely.
Will there be six leap seconds in 3000? Unlikely. But suppose you owned a very accurate wrist watch and
In a million years people might start adjusting the orbital speed of the earth to fix this. Or they might redefine the second to be a few more cesium transitions.
Personally I think it's a little far off to guess what we are going to be doing.
I have calculated[1] that in 1000 years a leap second will be required about every two months. It's likely that at that time we would still be using time standards similar to those in use now.
You're off by a bit, and are making some invalid assumptions to start with.
Steve Allen of the Lick Observatory has a great paper explaining the the fundamental clock problem [ucolick.org] and also exploring effects and impacts on society. It's really quite fascinating, and considerably more complex than most people imagine. I'v
The only problem is that no one knows how its supposed to work.
They work much like warp speed - start out low and end high. The higher the epsiode number, the higher the range. I think that it's derived from fishing, where "the big one that got away" gets bigger each time the story is told.
The other problem is that if starship travel at warp speeds, you run into problems with relativity, unless everyone uses one central clock. And if warp speed is >= (c) speed of light, then you're completely screwed.
Actually, and it's been a long time for me here, but I think Warp 1 = speed of light and the other warps went up from that exponentially. So warp 9.9 wasn't 9.9 times faster than the speed of light. It was many many many times greater than that...which is how they were able to zip around from one end of the quadrant to the other in a manner of days.
And of course, they explained away relativity effects because they were in a "warp bubble" when at warp speed...that is if they weren't trapped on the Holodeck
Note: Under the old system (used in the Original Series,) devised by Zephram Cochrane, the speed at any warp factor was given as: Ship's Velocity diveded by C is equal to the Warp Factor Cubed.
Therefore, in the Original Series, a Warp Factor of 3 was 27 times the speed of light, while a Warp Factor of 14 ws 2744 times the speed of light, or between Warp 9.6 and 9.9. I do not have the forumala for determining Warp Factors for the system, nor for the new system above 9. (They are the same formula, though
I think warp 10 can be reached by Douglas Adams' improbability drive. Or some other device that can be everywhere at the same time... I can't remember exactly, but it seems i can remember reading a story about something like this.
Yes, the nice thing about the Improbability Drive is that the ship occupies all points in space at the same time, and then appears in the place you want to be.
I don't care. I have an atomic watch that will automatically adjust. It is solar powered too, so I never have to worry about the bateries running down either. I love my watch.
3 hrs flourecent light = 1 hour sunlight. The watch will recharge indoors if nessecary. Plus, I work in the service industry. I am outside about 5 hours a day, every day during work hours.
Its always bugged me how in a leap year, we have an extra day... but a leap second is an entire extra second... if the terminology were consistent, that would mean a leap yer would equal one extra year.. or that a leap second was some fraction of a second longer than a normal one.
Yes, but by that logic, a leap anything added to a year should make that a leap year.. so is this a leap year because we are adding an extra second as other years are leap years because we add a day to it?
"To authorities responsible for the measurement and distribution of time..." How do I make an official complaint with these time-distributing authorities for all the times I've been blamed for being late? Everyone always thought it was my fault for running out of time--but nope, turns out there are "authorities" in charge of all that. Are they any relation to the Tooth Fairy?
Isn't this sort of thing calculable farther in advance? There shouldn't be a whole lot of angular momentum being added or subtracted from the Earth's rotation.
Do I need a new glibc? Or any other POSIX library, for that matter? If this is a new announcement then presumably every implementation of mktime(), localtime(), gmtime(), etc. needs to be updated.
Actually you'd just set the NTP server you're syncing with one second ahead. I don't think that a once-every-seven-year event is going to be worthy of kernel changes, especially since your time is probably off by more than one second as it is.
Actually you'd just set the NTP server you're syncing with one second ahead.
That simple solution will cause one of two problems: either your time will be off by one second, or the calculation of the number of seconds between two events crossing the leap second will be off by one second.
I don't think that a once-every-seven-year event is going to be worthy of kernel changes, especially since your time is probably off by more than one second as it is.
If you're regularly using NTPD your time is probably
If you're regularly using NTPD your time is probably not off by more than one second, and if you've got a GPS receiver it almost certainly isn't. But I agree with you that a kernel change isn't necessary, in fact, the kernel shouldn't be involved with terrestial time at all.
PC hardware sucks, or just mine. I sync ntpd every 4 hours on my workstation which is on all the time, consistantly it needs to adjust it by "0.31xxx" seconds. And that's just for four hours, imagine running it only once per day or on
Sounds like a software issue, not a hardware one. If your clock consistently needs to be adjusted by 0.31xxx seconds, then ntpd should be automatically adjusting by 0.31 seconds every 4 hours. ntpd will do this if it is run in daemon mode.
Actually, it's not calculable farther in advance. The Earth's rotation is inconsistent enough that leap seconds are sometimes needed, but the need can't be predicted more than about a year in advance [1]. In other words, there is noise in the Earth's rotation period of about 1 second per year. Atomic clocks are a lot better than this (good to ~50 ns per year [2]!!!), so it's pretty easy to detect the problem.
According to a link I just read, POSIX doesn't handle leap seconds [cr.yp.to]. So yes, if you use NTP, like someone else suggested, your time will be correct, but any measurements of time crossing leap seconds won't.
The correct solution in my opinion would be to store leap seconds along with the timezone information. That's really what they are. Unix time could be stored in TAI instead of UTC, and thus subtracting two times from each other would still give the correct result.
Well, regarding the second point, think of this as a reminder of why you should NEVER EVER store future time and date values as offsets from an epoch time/date, if being one second off is bad for your application.
Seriously, you'd think they'd pick a less conspicuous second to make the switch. Instead they picked the one second of the year that people notice more than any other. At least it'll only affect the New Year's count for the Brits. Here in the US we won't even be heading to the party yet.
IIRC, time is a key element of GPS accuracy. Even if updates come from atomic clocks, a human will have to key in a correction to a significant number of servers. Maybe Murphy and Chicken Little will show up for our New Years party.
Interesting. So we have "time in a bottle", so to speak.
Lets look at how time has progressed:
CT based on UTC by the ICWM at the 10th GCWM ES based on NTS ratified at the 11th, GCWM SI equals ES from USNO and NPL discovery as ratified at the 13th GCWM UT1 monitored by SRSPEOP of the IERS, thus we get LS adjustments according to CCIR.
Now we get TAI by BIPM expressed as UTC+dAT GPS is sychronized to UTC but not adjusted with LSs
OK, hold onto your pants... Until 1960, UT independant of AE. UT then replaced by ET. E
The last time I listened to WWV [slashdot.org] (the shortwave broadcast of NIST time), the signal said we were -0.6s ahead of time (ahead of the Earth's rotation). My thinking is that if we put in a leapsecond now, then we'd be -1.6s ahead - am I missing the point of the time broadcast code, or is this a hoax?
Microsoft has $33.8 billion in short term investments. Since interest payments are calculated by the day, and not the second, at an interest rate of 3% Microsoft will lose $1929 in interest due to this leap second.
12:59 on Dec. 31st is the only time in the whole year that people are paying attention to the seconds. Why not choose some other time when it's not going to screw up my countdown?
Any circuit design must contain at least one part which is obsolete, two parts
which are unobtainable, and three parts which are still under development.
Oh the opportunity! (Score:4, Funny)
Re:Oh the opportunity! (Score:2)
Re:Oh the opportunity! (Score:1)
Re:These guys must not be real nerds (Score:2)
Re:These guys must not be real nerds (Score:3, Funny)
Re:These guys must not be real nerds (Score:5, Interesting)
Re:These guys must not be real nerds (Score:2)
so 41 not divisible by 3
19 * 42 = 798
41 * 69 = 2829
------------------
3627 seconds / hour
60.45 minutes / hour
Nope, not perfect. Means I have to work an extra 3.6 minutes / day.
Re:These guys must not be real nerds (Score:2, Funny)
I think you need a leap integer in there somewhere.
Re:These guys must not be real nerds (Score:2)
1 / 3 =
2 / 3 =
57 / 3 = 19 : (42)
58 / 3 = 19.333 : (69)
59 / 3 = 19.666 : (69)
60 * 60 = 3600 : (69)
20 * 42 + 40 * 69 = 3600
Looks like it works to me.
Should we really bother? (Score:1)
Re:Should we really bother? (Score:5, Interesting)
I have calculated[1] that in 1000 years a leap second will be required about every two months. It's likely that at that time we would still be using time standards similar to those in use now.
On the other hand, in 1 million years, about 15 leap seconds will be required each day. Therefore, at some point timekeeping must necessarily divide the day into units that are not an integral number of seconds. We would have a situation where the record for the 100 metres dash is expressed in seconds, but the length of the second used for dividing up the day is not the same length. Such "stretched time" has already been used for the Spirit and Opportunity rovers on Mars.
[1] A common formula for approximating the evolution of delta-T over time is 31 * Cy^2, where Cy is expressed in centuries.
Re:Should we really bother? (Score:3, Interesting)
Re:Should we really bother? (Score:2)
A standardized second. (Score:3, Funny)
Lifted from wikipedia - The second is one of seven SI base units. It is defined as 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 at zero kelvins.
Now that we have a definition perhaps someone could tell us what it means.
Re:A standardized second. (Score:2, Informative)
As another example, take the metre. You can see how the definition of the metre became more and more precise here [wikipedia.org]. I don't see it mentioned there, but the original "meaningful" definition of the metre, if extremely imprecise, is based on water: 1 ml of water = 1 gm in mass =
Re:A standardized second. (Score:2)
The mass definition was thus based on the length of the metre, not the other way around.
Re:A standardized second. (Score:1)
Re:A standardized second. (Score:2)
Re:A standardized second. (Score:2)
The definition is made in terms of the most accurate way we have of measuring it - with the atomic clock. There's a description of how they work here [howstuffworks.com].
In the future the definition may change - there are developments to produce "optical clocks" which are more accurate
Re:Should we really bother? (Score:1)
You dont think things will change in 1000 years? It's hard to keep things the same for 10.
And that will just get more tedious once we abandon the "year", of course.
Re:Should we really bother? (Score:3, Interesting)
Simple solution, change the 60 Hz power line frequency to 59.9999885922 Hz, causing power-line synchronous clocks to slow down to match the daily rotation of the Earth (as referenced to the Sun).
On the other hand, in 1 million years, about 15 leap seconds will be required each day.
Simple solution, change the 59.9
Re:Should we really bother? (Score:2)
It's very close to the engineered value (50 or 60) but is constantly fluctuating.
That is why AC powered clocks that use the HZ to drive them tend to need adjustment over time.
Re:Should we really bother? (Score:2)
They'll just adjust the placement of the leap year, which is much simpler than doing constant leap second adjustments.
Dunno if that's an attempt to get a funny mod or what, but I'll take it seriously.
There's only one slight problem with this, noontime as the clock will happen an hour and a half earlier in the day every year, until after eight years noon occurs in the middle of the night. After another eight years of this silliness, you can add an extra Leap Da
Re:Should we really bother? (Score:2)
The formulae are to some extent empirical as well as being approximations. The evolution of delta-t is also extremely 'noisy', and is far from a good fit with any of the formulae [phys.uu.nl].
For the last few years it's been at around 64 seconds (see data in here [navy.mil]), which comes to about 3 or 4 seconds less for the present time than some recent formulae were predicting even only a few years ago, and l
Re:Should we really bother? (Score:2)
This is why I made sure to state that the formula was an approximation.
Although it is unlikely that such formulae are going to be accurate 1 million years form now, it does not detract from the point that I raise, namely that our current definition of 86400 seconds to the day and 1 second = 9192631770 cesium transitions cannot both last indefinitely.
Will there be six leap seconds in 3000? Unlikely. But suppose you owned a very accurate wrist watch and
Re:Should we really bother? (Score:2)
Personally I think it's a little far off to guess what we are going to be doing.
Re:Should we really bother? (Score:2)
You're off by a bit, and are making some invalid assumptions to start with.
Steve Allen of the Lick Observatory has a great paper explaining the the fundamental clock problem [ucolick.org] and also exploring effects and impacts on society. It's really quite fascinating, and considerably more complex than most people imagine. I'v
Yes (Score:1)
I remeber reading about some Unix library or whatnot (ctime or something) that could handle them... musta been a slow news day on
Star Trek has it figured out. (Score:4, Funny)
The only problem is that no one knows how its supposed to work.
Re:Star Trek has it figured out. (Score:5, Interesting)
They work much like warp speed - start out low and end high. The higher the epsiode number, the higher the range. I think that it's derived from fishing, where "the big one that got away" gets bigger each time the story is told.
Relativity (Score:1)
Re:Relativity (Score:2)
Re:Relativity (Score:2)
And of course, they explained away relativity effects because they were in a "warp bubble" when at warp speed...that is if they weren't trapped on the Holodeck
Re:Relativity (Score:2)
From the Star Trek Encyclopedia:
Note: New Cruising Speed
Note: Old Cruising Speed
Note: Old Normal Maximum Speed
Maximum rated Speed, can be maintained for 12 hours
Auto-shutdown of engines after 1
Re:Relativity (Score:2)
Note: Under the old system (used in the Original Series,) devised by Zephram Cochrane, the speed at any warp factor was given as: Ship's Velocity diveded by C is equal to the Warp Factor Cubed.
Therefore, in the Original Series, a Warp Factor of 3 was 27 times the speed of light, while a Warp Factor of 14 ws 2744 times the speed of light, or between Warp 9.6 and 9.9. I do not have the forumala for determining Warp Factors for the system, nor for the new system above 9. (They are the same formula, though
Re:Relativity (Score:2)
Re:Relativity (Score:2)
Yes, the nice thing about the Improbability Drive is that the ship occupies all points in space at the same time, and then appears in the place you want to be.
Generally speaking, that is.
Shouldn't that be... (Score:2)
Re:Shouldn't that be... (Score:5, Informative)
The sequence of dates of the UTC second markers will be:
2005 December 31, 23h 59m 59s
2005 December 31, 23h 59m 60s
2006 January 1, 0h 0m 0s
Actually, its 12:00:00 then another 12:00:00.
I can see it now... (Score:1, Funny)
HAPPY NEW YEAR!!!
Re:Shouldn't that be... (Score:2)
Re:Shouldn't that be... (Score:2)
Re:Shouldn't that be... (Score:2)
Good thing neither the editors or news submitters can tell time properly! (23h 59m 59s = 11:59:59 PM, 0h 0m 0s = 12:00:00 AM)
Re:Shouldn't that be... (Score:3, Informative)
0h 0m 0s = Midnight
12h 0m 0s = Noon.
These is no such time as 12:00:00am (or 12:00:00pm).
See NIST [nist.gov] for the gory details.
Yay! (Score:3, Funny)
I have an atomic watch! (Score:1)
Re:I have an atomic watch! (Score:2, Funny)
Re:I have an atomic watch! (Score:1)
Re:I have an atomic watch! (Score:3, Informative)
Re:I have an atomic watch! (Score:2)
Re:I have an atomic watch! (Score:1)
Re:I have an atomic watch! (Score:1)
Its always bugged me how... (Score:3, Interesting)
Re:Its always bugged me how... (Score:2)
Re:Its always bugged me how... (Score:1)
Re:Its always bugged me how... (Score:2)
Re:Its always bugged me how... (Score:1)
Re:Its always bugged me how... (Score:2)
Tire Rotation (Score:2, Funny)
Welcome to the International Earth Rotation and Reference Systems Service
Will they change my oil and check my brakes, too?
I knew it! (Score:3, Funny)
Two questions (Score:3, Insightful)
Do I need a new glibc? Or any other POSIX library, for that matter? If this is a new announcement then presumably every implementation of mktime(), localtime(), gmtime(), etc. needs to be updated.
Re:Two questions (Score:3, Insightful)
Re:Two questions (Score:2)
Actually you'd just set the NTP server you're syncing with one second ahead.
That simple solution will cause one of two problems: either your time will be off by one second, or the calculation of the number of seconds between two events crossing the leap second will be off by one second.
I don't think that a once-every-seven-year event is going to be worthy of kernel changes, especially since your time is probably off by more than one second as it is.
If you're regularly using NTPD your time is probably
Re:Two questions (Score:1)
PC hardware sucks, or just mine. I sync ntpd every 4 hours on my workstation which is on all the time, consistantly it needs to adjust it by "0.31xxx" seconds. And that's just for four hours, imagine running it only once per day or on
Re:Two questions (Score:2)
Re:Two questions (Score:5, Informative)
Actually, it's not calculable farther in advance. The Earth's rotation is inconsistent enough that leap seconds are sometimes needed, but the need can't be predicted more than about a year in advance [1]. In other words, there is noise in the Earth's rotation period of about 1 second per year. Atomic clocks are a lot better than this (good to ~50 ns per year [2]!!!), so it's pretty easy to detect the problem.
Sorry, I can't help with the second question.
[1] See this Wikipedia article [wikipedia.org].
[2] See this Wikipedia figure [wikipedia.org].
Re:Two questions (Score:1)
Re:Two questions (Score:3, Informative)
According to a link I just read, POSIX doesn't handle leap seconds [cr.yp.to]. So yes, if you use NTP, like someone else suggested, your time will be correct, but any measurements of time crossing leap seconds won't.
The correct solution in my opinion would be to store leap seconds along with the timezone information. That's really what they are. Unix time could be stored in TAI instead of UTC, and thus subtracting two times from each other would still give the correct result.
Whenever a leap second was announced y
Re:Two questions (Score:2)
Off by one error (Score:4, Interesting)
Re:Off by one error (Score:2)
NINE!
EIGHT!
SEVEN!
SIX!
FIVE!
FOUR!
TWO!
ONE!
ONE!
HAPPY NEW YEAR!
lameness filter encountered reason: don't use so many caps. it's like yelling.
-
Re:Off by one error (Score:1)
Re:Off by one error (Score:4, Insightful)
Ugh! (Score:2, Funny)
Re:Ugh! (Score:1)
lol
12:59 AM or PM? (Score:3, Informative)
leapsecond.com (Score:4, Informative)
Okay, here's a clickable link:
http://leapsecond.com/ [leapsecond.com]
An obsession in another are of time is this Y10K Compliant clock:
http://longnow.org/ [longnow.org]
How will GPS be affected? (Score:1)
Even if updates come from atomic clocks, a human will have to key in a correction to a significant number of servers.
Maybe Murphy and Chicken Little will show up for our New Years party.
Re:How will GPS be affected? (Score:2)
Re:How will GPS be affected? (Score:1)
So we have "time in a bottle", so to speak.
Lets look at how time has progressed:
CT based on UTC by the ICWM at the 10th GCWM
ES based on NTS ratified at the 11th, GCWM
SI equals ES from USNO and NPL discovery as ratified at the 13th GCWM
UT1 monitored by SRSPEOP of the IERS, thus we get LS adjustments according to CCIR.
Now we get TAI by BIPM expressed as UTC+dAT
GPS is sychronized to UTC but not adjusted with LSs
OK, hold onto your pants...
Until 1960, UT independant of AE.
UT then replaced by ET.
E
58 (Score:1, Redundant)
I thought we were behind (Score:1)
Bad for Microsoft (Score:4, Funny)
why dec 31st? (Score:1)
Correction (Score:2)