The current UTC standard permits a lot of flexibility in scheduling leap seconds. A leap second, either positive or negative, is permitted at the end of each month. Current policy limits leap seconds to only occur at the end of June or December, but the international standard on the books (that all conforming usage should agree with) allows every month.
If systems have problems handling the rare occurrence of leap seconds (and would have vastly larger problems handling the very rare leap hours), one might suggest that permitting opportunities for leap seconds *more frequently* would serve to convince recalcitrant project managers to implement the standard they already are obligated to honor.
One might argue (as I have) that we simply allow leap seconds to be issued every month. Among other things, this scheduling freedom would reduce the RMS error between UTC and UT1. It would certainly be much harder for technical projects to ignore the actual requirements of civil time through ignorance.
Another suggestion (can't take credit for this one) would make leap second scheduling completely predictable (in a sense). Simply schedule a leap second at the end of every month, alternating positive and negative. Thus an actual leap second required by our inconstant orb would become the omission of an expected event.
- The leap year algorithm can be static because the calendar changes slowly. The corresponding algorithm for time is the 24 hour clock itself, not leap seconds, which are a higher order effect whose equivalent is currently completely ignored (as it should be) for the calendar.
- We don't have leap seconds because the Earth is transferring angular momentum to the lunar orbit. We have leap seconds because the Earth has already slowed down since the 1820 epoch that effectively corresponds to the definition of the SI second.
- We have just passed through a seven year lull in leap seconds. The overall trend is one leap second per 18 months, corresponding to a 2 ms/day mismatch between atomic time and the spinning Earth.
- The length of the mean solar day is growing by about an additional 2 ms per century. As the length of day grows, leap seconds will be required more and more frequently - a quadratic effect. Don't panic, though! The current UTC standard is good for hundreds of years yet.
- The real issue isn't the variability of the Earth's rotation, it is the misidentification of the original definition of the second as 1/86400 of a (varying) mean solar day with the more recent (static) SI unit of time. The original proposal was to call the SI unit an "essen" in honor of a well known (to some) timekeeper.
- When the mean solar day is increased by one full second (hundreds of centuries hence), the moon will have receded by about one mile.
- The day is actually *shorter* now than a hundred years ago, because the long term lunar trend is dwarfed by short term variations that are an order of magnitude larger (see http://www.ucolick.org/~sla/leapsecs/dutc.html).
- Each missed leap second would be equivalent to moving the prime meridian 300 meters at the latitude of Greenwich.
Slashdot Mirror
The current UTC standard permits a lot of flexibility in scheduling leap seconds. A leap second, either positive or negative, is permitted at the end of each month. Current policy limits leap seconds to only occur at the end of June or December, but the international standard on the books (that all conforming usage should agree with) allows every month.
If systems have problems handling the rare occurrence of leap seconds (and would have vastly larger problems handling the very rare leap hours), one might suggest that permitting opportunities for leap seconds *more frequently* would serve to convince recalcitrant project managers to implement the standard they already are obligated to honor.
One might argue (as I have) that we simply allow leap seconds to be issued every month. Among other things, this scheduling freedom would reduce the RMS error between UTC and UT1. It would certainly be much harder for technical projects to ignore the actual requirements of civil time through ignorance.
Another suggestion (can't take credit for this one) would make leap second scheduling completely predictable (in a sense). Simply schedule a leap second at the end of every month, alternating positive and negative. Thus an actual leap second required by our inconstant orb would become the omission of an expected event.
Fun facts about civil timekeeping:
- The leap year algorithm can be static because the calendar changes slowly. The corresponding algorithm for time is the 24 hour clock itself, not leap seconds, which are a higher order effect whose equivalent is currently completely ignored (as it should be) for the calendar.
- We don't have leap seconds because the Earth is transferring angular momentum to the lunar orbit. We have leap seconds because the Earth has already slowed down since the 1820 epoch that effectively corresponds to the definition of the SI second.
- We have just passed through a seven year lull in leap seconds. The overall trend is one leap second per 18 months, corresponding to a 2 ms/day mismatch between atomic time and the spinning Earth.
- The length of the mean solar day is growing by about an additional 2 ms per century. As the length of day grows, leap seconds will be required more and more frequently - a quadratic effect. Don't panic, though! The current UTC standard is good for hundreds of years yet.
- The real issue isn't the variability of the Earth's rotation, it is the misidentification of the original definition of the second as 1/86400 of a (varying) mean solar day with the more recent (static) SI unit of time. The original proposal was to call the SI unit an "essen" in honor of a well known (to some) timekeeper.
- When the mean solar day is increased by one full second (hundreds of centuries hence), the moon will have receded by about one mile.
- The day is actually *shorter* now than a hundred years ago, because the long term lunar trend is dwarfed by short term variations that are an order of magnitude larger (see http://www.ucolick.org/~sla/leapsecs/dutc.html).
- Each missed leap second would be equivalent to moving the prime meridian 300 meters at the latitude of Greenwich.