[LEAPSECS] The relation between calendars and leap seconds.

[Rob Seaman]

John Cowan wrote:


> What has the Moon to do with it?  The connection of the Moon to the

> calendar was lost in Julius Caesar's time -- doubtless to great  

> howling

> by the astro{nom,log}ical community.


...and around we go again.  This was my shorthand way of referring to  
all the issues associated with variations in Earth orientation, in  
particular the secular trend leading to the quadratic behavior.  As a  
reminder, it has long since been shown that leap seconds would be  
required even in the absence of tidal deceleration.

The issue with the calendar, is, has been, and always will be that a  
clock is a subdivision of a calendar.  You can't muck around with leap  
seconds without calling into question the definition of the day.

It is a false equivalence to point to the multitude of cyclic  
celestial phenomena, and draw a conclusion that none matter because  
some are of little importance to our daily lives.  Just examine that  
phrase - daily lives.  Days matter in civil timekeeping.  Fractions of  
days matter.  Secular drift of days matter.

Every scheme proposed in the original GPS World article amount to sly  
cheats on mean solar time - including the eradication of leap  
seconds.  This is not an option that could possibly work if the SI  
second were much more different from the mean solar second (1/86400 of  
a solar day) than it is now.  The only reason we're having this  
discussion is that the paleo timelords didn't stick to the plan to  
call the SI unit an "essen" instead of a "second", thus avoiding the  
whole confusion in the first place.

Adi Stav wrote:


> We don't how many seconds there will be in 2009-12-31 23:59; there

> might be 60, 59, or 61. So the UTC calendar is not predictable.

>

> I can easily imagine such predictable systems, such as adding  

> "permenant

> leap seconds" regularly into the year according to some pre-determined

> formula, but I don't think I've every seen such a thing proposed.


We've thrashed out seven (no, not literally seven) different  
variations over the unending aeons of this discussion.  There is  
absolutely nothing in the current definition of UTC to stop us from  
announcing a schedule years in advance.  This is by far the easiest  
change to make to UTC.

If we weren't spending all our time trying to fend off the unilateral  
cessation of leap seconds, we could put that energy into discussing a  
new scheduling algorithm.  Not permanent, perhaps, but some other  
trade-off between amplitude and scheduling horizon seems eminently  
achievable.

UTC is not perfectly predictable because the Earth is not.   
Emasculating UTC won't make the Earth any more predictable, or human  
civilization any less synchronized with diurnal cycles.

There are options yet unexplored.  We'll get to them faster by  
focusing on requirements than by reiterating our talking points over  
and over and over again.

Rob