[LEAPSECS] Metrologia on time

[Warner Losh]

No wonder it just got cold here.. <g>

Seriously, I think that nobody would say that more advance notice of leap seconds is bad.

I'd go one step further, assuming the current elimination plans fail, and say they should be announce 10 years out, but that DUT1 can drift up to 2.0s or maybe a little more,  but in the long haul it will be kept in sync.  It is what we do with leap years today: nobody cares that for a century or two we may be a day or a little more off.  We correct that by not having a leap year every 400 years which puts us on a good footing for several thousand years, by which time LOD changes will start to foul that calendar up...  The long term average is good, even if the short term cycles aren't perfect.

Warner

On Aug 2, 2011, at 11:53 AM, Rob Seaman wrote:


> PHK and I are on the same page with this.  By all means announce leap seconds with significantly more notice.  - Rob

> --

> 

> On Aug 2, 2011, at 10:39 AM, Poul-Henning Kamp wrote:

> 

>> ...

>> I find it much more intesting that Bernard Guinot goes on the record

>> on page 183:

>> 

>> 	They are not theoretically predictable, but smooth enough

>> 	for an empirical prediction to less than ±1 s over 3 years

>> 

>> and:

>> 	Tests on real data from 1955 to the present, made by the

>> 	author, have shown that a linear prediction based on one

>> 	year of values of UT1 - TAI led to a maximum error less

>> 	than 0.6 s in two years after the last observed values.

>> 	After three years, the maximum was 1.0 s.

>> 

>> So giving us 3 years notice of leap seconds instead of six months

>> should be a total no-brainer.

> 

> _______________________________________________

> LEAPSECS mailing list

> LEAPSECS at leapsecond.com

> http://six.pairlist.net/mailman/listinfo/leapsecs

> 

>