[LEAPSECS] Unix Questions

[Rob Seaman]

On Aug 6, 2010, at 2:00 PM, M. Warner Losh wrote:


> We have about 1 leap second every 18 months now, and that rate is  

> basically accelerating.


The divergence between the length of the actual second (1/86400 of a  
solar day) and the SI second (a static unit of the metric system) will  
continue accelerating whether there are leap seconds or not.  The  
Earth is a top being slowed by the tides.

We would have been saved a lot of grief if the SI second had been  
given a distinct name.  It is the confusion between the two very  
different meanings of the word "second" that has led to this issue  
even being considered to be an issue in the first place.

Maybe we can convince Home Depot to start selling lumber labeled  
according to its "plank's constant".

There are two kinds of clocks - atomic and solar.  Pretending  
otherwise won't make this statement untrue.


> A leap hour would be needed in several hundred years.


A "leap hour", unlike a leap second, is an undefined concept.  It is  
quite different, for instance, from the spring forward / fall back  
hours of daylight saving adjustments.  As with a leap second it would  
presumably occur between the same two clock ticks worldwide.  Unlike a  
leap second, however, this intercalary event would have to be wedged  
in as something like the 3:00 pm hour - all over again.  By contrast,  
a DST adjustment occurs at some out of the way hour like 2:00 am on a  
Sunday - local time.  The change would affect the underlying legal  
timescale.  Localities are free to play wacky games with DST precisely  
because UTC goes clicking along as usual.  The discontinuity in the  
timeline of history would be unlike anything that has ever come before.

While waiting for the unplanned and undefined leap hour to occur, our  
clocks would run more and more fast relative to the Earth.  It is this  
perpetually uncorrected rate error we should be discussing, not the  
mythical leap hour.


> I think the plan is to ditch leap seconds,


The issue is precisely that there is no publicly accessible plan.   
 From what we've been able to glean about the proposal it is clear  
that there is no coherent plan in private either.  It is not  
sufficient to cease issuing leap seconds.  There must be some hint of  
a plan for what will happen when the divergence between the clocks and  
the real world inevitably reaches a scale that can't be ignored.

For astronomers that point will be reached immediately.  We are the  
canaries in the coal mine.

What is the plan for distributing the ever more quickly increasing  
(and ever more critical to get right) DUT1 correction whose large  
magnitude will rapidly exceed the current limit of 0.9s?  The argument  
(to the extent that there is a coherent argument) is precisely that  
"nobody cares" about leap seconds.  They don't care precisely because  
DUT1 is kept small and thus negligible for most purposes.

Rob Seaman
NOAO