NavList:

Geoffrey Kolbe, you wrote:
"However, allowing UTC to drift without leap-second corrections will mean that the time will come when we are getting up in the middle of the night, having lunch when dawn is breaking and going to bed when the sun is still high in the sky!"

Yes, but that's a very long time in the future. By current estimates, even in 2000 years the accumulated time offset would amount to about four hours. It's true that this is an eventual concern, and we have to worry even today about populist concern that we might at some point screw up the days of the week. Shifting Sunday to Monday would be a bad thing in public relations terms, even if that happens thousands of years in the future on some theoretical calendar. But let's be clear; this is far off in the future. On a more immediate timescale, if leap seconds were dropped in the year 2020 (not going to happen, but just for the sake of argument), the accumulated offset of UT1-UTx (UTx being UTC without leap seconds) would be about one minute by the year 2070, about five minutes by 2170, and would not reach 15 minutes until the year 2370 ...more than a quarter of a millennium from now.

And:
"In some way, civil time has to be maintained as a reasonable match the observed solar day and some mechanism needs to be in place to ensure that this will happen. At the moment it is a relatively small correction that happens once or twice a year, which is often enough that the ability to take the correction into account is built into all the electronic instruments that need to take it into account."

I can't agree with that last bit. Many systems use the famous Google smear where they run their system clocks at a slower rate for a day to generate the equivalent of a leap second by the end of the day. This is, at best, a hack --a trick to work around the issue, rather an implementation of a solution. And a great many devices simply fail to incorporate the leap second in any way and have to be reset or re-synchronized at some point after the actual time of the leap second insertion. It's a mess. The big problem, of course, is that the insertions are not algorithmic. They are arbitrary events, declared by a small organization. They've done a good job at this over the decades, but it's still an arbitrary activity. What can we do in the future?

Continuing with the scenario above, if leap seconds are abolished, then there are a number of options which could come into play around the year 2600 when the time offset has reached 30 minutes. The simplest is to use the pre-existing "daylight time" (summer time) rules, but with the addition that they will be applied once everywhere and not just in the mid-latitude countries that currently use daylight time. We agree that in the year 2599, those countries which normally add one hour in Spring will skip that clock advancement (just once) and countries which do not use daylight time turn their clocks back one hour at some date close to the "Fall back" date used in the northern hemisphere. Here I am assuming that the southern hemisphere is still an insignificant backwater economically in the year 2599, but that could change ;). Obviously this will work better if there is some international synchronization, but that's merely a technical, legal detail. It will be resolved by future legislation ...and future legislators. Or maybe those future legislators will decide that individual countries have the sovereign right to do as they please. Wouldn't that be crazy?

You concluded:
"Any other correction will be yet-to-be-agreed, large, and happen with such irregularity that it will be bad for all sorts of other reasons other than navigation!"

Could we make the same argument about the 400-year rule on leap years? Maybe we should? The biggest difference between that yet-to-be-agreed correction and the current system is that it can be algorithmic for up to a thousand years into the future. Once some agreement is made, it will occur according to known rules. Of course, individual countries can and do change their time zone assignments and the time zone assignments of regions and states within countries. The idea that shifting our clocks by some fixed amount, like one hour, would be highly disruptive seems to be contradicted by our experience with daylight time and with these time zone re-assignments.

Residents of major western countries have been off the local solar time standard in our lives since the late 18th/early 19th centuries when time by mechanical clocks and watches became standard. That evolution disconnected us from Sun time by +/- 15 minutes during the year (exact value equal to the "equation of time" for any date). In the late 19th century, up to an additional half an hour was added as countries adopted standard time zones. This was quickly increased to an hour and sometimes more as regions (like the eastern half of the American Midwest) sought to align their time with early rising business centers (moving Ohio, Michigan, and most of Indiana to New York time). And in the 20th century, daylight time added another hour of offset, variable by season, to our separation from Sun time. By now. many locations are comfortably as much as two hours removed from solar time during their normal civil time-keeping day. Do they notice? Are the lives of people in Kalamazoo, Michigan disrupted because they keep their clocks set, for most of the year, to a time that would be appropriate, in terms of mean solar time for Barbados and Trinidad? Maybe a little. Yet clearly this evidence demonstrates that our daily lives can indeed be disconnected from solar time by considerably more than one hour with little downside. And we can jump our clocks by an hour twice a year without calamity. Surely at the end of the 26th century we can expect to do the same. A one hour shift in clocks then, to cancel out the 30-minute cumulative total of missed leap seconds by that date, should cause no significant problem.

Finally, just to reiterate for others following along, it is certainly not the case, as far too many celestial navigation enthusiasts believe, that dropping leap seconds would somehow "break" traditional celestial navigation. The correction would be a relatively simple time offset. It is already the case that the input time variable for the almanac is UT1, and for utmost accuracy any UTC value should be adjusted by some tenths of a second before interpolating in the ephemerides. If leap seconds are dropped, that adjustment (now an adjustment to UTx) becomes some larger number of seconds. For example, in the year 2150, a future Nautical Almanac, might include the instructions "Subtract 4m18s from UTx before Mar 3, subtract 4m19s from Mar 4 to Sep 27, subtract 4m20s after Sep 28. Of course, since the almanacs are only published some months before the beginning of the year even now, the correction could also be included directly in the ephemerides so that the entry could be UTx to the nearest second, just as it is now. It would also be possible to arrange for navigators to carry "chronometers" (virtual or otherwise) set and rated to UT1. As long as the rules are simple and clearly stated, there is no problem.

Frank Reed
ReedNavigation.com