[LEAPSECS] Short notice for DST changes

Rob Seaman seaman at noao.edu
Tue Feb 25 09:24:14 EST 2014

On Feb 25, 2014, at 12:10 AM, Ian Batten <igb at batten.eu.org> wrote:

> On 24 Feb 2014, at 22:56, Rob Seaman <seaman at noao.edu> wrote:


>> Rather, there have to be very close to 86,400 SI-seconds per day, because that how long a day is.


> And this is why engineers and scientists make bad public policy, because of an unwillingness to make convenient but "unscientific" approximations.

This is nonsense of course. Nobody knows better than scientists the practice of making approximations; and the essence of engineering is the art of trading off one figure of merit against another. Every trade-off study is coupled with sensitivity analyses in various metrics.

The argument also cuts both ways. The ITU-R includes a plethora of engineers and scientists.

> All the arguments about rate comes down is what the definition of "very close" is.

No, the argument comes down to the fact that there has been no argument. We've been talking for 15 years and precious few of those pushing this agenda have deigned to communicate with this list or with other fora on the topic.

And the agenda has never been couched in terms of approximation, but rather in terms of simply ceasing leap seconds so as to turn UTC into a purely atomic timescale. One problem with this is indeed that ultimately the embargoed leap seconds will have to reappear somewhere.

A more fundamental problem is that access to purely atomic timescales is already straightforward, either TAI computed from UTC - or simply via GPS. The proposal isn't about improving access to atomic time, it's precisely about denying access to solar time.

> A day isn't 86400 SI seconds, nor is the average day 86400 SI seconds. Hence, whatever you do, there will need to be corrections made in some manner over a sufficient long time-scale.

Glad to hear your acknowledgement of this...

...with the addendum that it must happen frequently enough that Coordinated Universal Time continues to function as Universal Time. Any timescale that doesn't should not include "Universal Time" in its name.

> There are for practical purposes no applications outside astronomy where you need alignment of clocks to synodic days to a precision of more than a hour,

That will come as a surprise to the American Astronautical Society who published our two proceedings:

http://www.univelt.com/Science.html **

I may be biased since my Dad was in aerospace, but communications and weather satellites have always seemed practical to me. GPS itself requires Universal Time to operate:


Again, this cuts both ways. Few communities place more stringent requirements on atomic time than astronomers.

> or more: we know this, because many large population centres have their local noon an hour or more from mean solar noon, [...]

This misidentification of requirements layered on mean solar time with anecdotal use cases from apparent solar time keeps cropping up like a weed. Timezone offsets are static (and should remain so). The equation of time and daylight saving time offsets are periodic. Leap seconds, however, are about compensating for a secular trend. Relative amplitude is immaterial to the functional form.

> for people who do need synodic time, the Internet can provide it.

For people who do need atomic time, the internet and GPS already provide it.

> ludicrous scenarios are proposed about disabled orphans who communicate only in ancient Egyptian and have religious objections to payment in anything other than gold coins.

I believe you're the only one to have mentioned these poor souls. Do tell?

> Similarly, arguing that the civil time should be vetoed by some tiny pool of astronomical equipment is not going to convince anyone outside the astronomical community, and not even all of those.

Rather, it is simply fact that day means synodic day and that its variable length will always have to be reconciled with the constant metronome of atomic time. There is indeed more than one way to consider doing this. The ITU proposal has never included any of these possibilities, but rather simply suggests that we pretend the engineering requirement does not exist, that leap seconds simply cease, that there is no distinction between the different timescales. It is an exercise in institutionally mandated public ignorance.

> So there is a massive approximation in use already,

No, UTC approximates mean solar time at Greenwich to within +/- 0.9 seconds. Mean solar time is not apparent solar time, nor is it atomic time. It is simple fact that these three things are different even though they all have the word "time" in them. One way or another these distinctions will make themselves known. This is the essence of any attempt to ignore an engineering requirement. It is called a requirement because it is required by the situation.

> I suspect that plans for 2014 made in 1514 have not proven to be durable.

How about 1582, when the Gregorian calendar was created? Or 46 BCE? The Julian calendar is still used for liturgical purposes and was employed for civil purposes into the 20th century. Of all the arguments made here, the least persuasive is that we shouldn't bother making plans for something as fundamental as timekeeping.

Rob Seaman
National Optical Astronomy Observatory

** Preprints at:

http://www.cacr.caltech.edu/futureofutc/2011/preprints/ and

Comments welcome.

More information about the LEAPSECS mailing list