25 o'clock, we'll be together 'til the end of time (was Re: [LEAPSECS] Longer leap second notice)

From: Rob Seaman <seaman_at_noao.edu>
Date: Wed, 4 Jan 2006 15:29:22 -0700

(Extra credit for those who recognize the quote.)

Tom Van Baak wrote:

>>> A more apt comparison would be to the leap year rules that we
>>> have. We know the rules going
>>> forward a thousand years or so.
>>
>> Apt indeed. Leap seconds are scheduled at least six months in
>> advance. That's about one part in 15 million. A thousand year
>> horizon for scheduling leap days is one part in 365,250. So we're
>> already doing 50 times better than Julius Caesar and Gregory XIII.
>
> I'm glad someone else came to this conclusion too.
> But be careful, though, that this line of thought isn't
> used to justify leap hours over leap seconds. One
> leap hour predicted for a fixed date of, say, the year
> 2600 is about one part in 5 million; still 15x better
> than the Gregorian calendar.

One of the more disquieting aspects of the leap hour
"proposal" (really a "notional position") is the tolerance for slop
in our clocks. If you don't give a rat's ass for enforcing any kind
of limit on the amplitude of the excursion that is more precise than
"an hour", you can indeed schedule a leap hour for any particular
decade or century that you care to name. This is especially true if
you don't even bother to discuss *how* the leap hour will be
implemented other than by loose analogy with daylight saving time.

On the other hand, if an explicit leap hour trigger condition were
specified (i.e., "issue a leap hour on the next December 31 after |
DUT1| exceeds 31 minutes"), it would be at least as difficult to
predictively schedule leap hours as leap seconds. One expects there
would be strong pressure to issue the leap hour (singular and unique
because I'm convinced one would be more than enough for civilization
to bear) - strong pressure, that is, to issue the leap hour at some
round dut1 boundary - 30, 40, 50 minutes. (I think some have been
thinking, "count up to 60:00 and subtract it off again".) In point
of fact, one might expect that the prospect of undergoing a leap hour
during a particular lifetime would be so daunting that the leap hour
would be delayed as long a possible. (That's the whole driving
motivation, after all.) This is directly counter to the evident
policy for issuing leap seconds that has historically resulted in
each leap second occurring as soon as practical (presumably to allow
"slack" the next time).

Also, a leap second is not implemented using the same method as
daylight saving time. A leap second is not a "fall back" second, but
rather a "do it twice" second - precisely to preserve the historical
flow of time. This would be a much more critical issue for leap hours.

There appears to be an assumption that a leap hour would be
implemented as an extra "fall back" hour inserted into the stream of
daylight saving. Let's ignore the fact that many localities do not
currently observer daylight saving and have no infrastructure for
implementing such. Let's also ignore the question of whether anybody
at all will observe daylight saving centuries hence. Rather let's
focus on how to maintain the historical flow of time across this
factor of 3600X higher temporal cliff. The reason daylight saving
can ignore this issue is precisely that all localities observe an
underlying standard time for which no discontinuity occurs. A
contract, for instance, can be specified with respect to standard
time and the implications of that contract are coherently
understandable to all parties even *during* a "spring forward" or
"fall back" jump.

But a leap hour is an adjustment of the underlying standard time.
Again, let's ignore the very real question of how all localities
might be convinced to implement a leap hour simultaneously. Without
a stable underlying reference clock, a standard time zone undergoing
a "fall back" leap hour would experience a colossal crack in time. I
assert that our mutated, gill-breathing, Christmas-hating great-
great-...-great-grandchildren would clearly not choose to implement a
leap hour as a "fall back" hour. Rather, they would adopt the "do it
twice" algorithm currently used for leap seconds. One speculates
that the day in question would simply be declared to be 25 hours
long. This would preserve the "historical record". (GalaxyQuest
anyone?)

Note however, how much more prominent a leap hour becomes. Nobody
has surely expected it to be something that virtually all of
civilization can ignore - like, say - a leap second. But unlike the
naive notion of simply "falling back" twice one year, this becomes
not just a big deal on the day in question - not just a big deal
during the year in question - not just a big deal during the century
in question - but would be a unique occurrence in all of recorded
human history. It would henceforth be the "25 Hour Day".

More than four centuries later, historians still have to keep track
of the calendar change. The English speaking world conflated the
trouble by delaying implementation for 170 years. How much more
confusing should some locales delay - perhaps guided by local
religious conviction or some active military or political conflict -
should some locales delay the implementation of a leap hour. Or
perhaps some might simply, naively, in their child-like innocence,
choose the divergent "fall back" method of carrying it out. Whaddamess!

(Note that I haven't even bothered to address the question of how you
might successfully interpolate an entire extra hour at different
times of day in different time zones around the world. Wrap your
head around that one.)

In a follow-up message, Tom strengthens my position:

> are there any concerns over the growing(?) use of double :59 second
> or double :00 second instead of :59:60 for a positive leap second?

In the case of leap seconds, such ambiguity is purely a technical
question for the experts. On the other hand, it would be absolutely
critical to avoid this ambiguity - worldwide - with leap hours. One
can only wonder what a peer review panel would make of the complete
silence of the ITU proposal on all such implementation "details".

Rob Seaman
National Optical Astronomy Observatory
Received on Wed Jan 04 2006 - 14:29:53 PST