[ntp:questions] Purpose of a leap second file?

unruh unruh at invalid.ca
Sun Mar 4 16:27:15 UTC 2012

On 2012-03-04, David E. Ross <nobody at nowhere.invalid> wrote:
> On 3/3/12 1:13 PM, John Hasler wrote:
>> David E. Ross writes:
>>> Why require a UTC adjustment from TAI?  If the data represent a
>>> satellite orbiting the earth and the sub-satellite point on the
>>> earth's surface is needed, the TAI is converted to UTC, which is then
>>> converted through UT2 to UT1 and possibly to UT0 before obtaining the
>>> sidereal time.  Sidereal time then gives the current angle of the
>>> earth's rotation for computing longitude.
>> Thus UTC is just a kind of timezone-like localization and leap-seconds
>> belong in tzdata.
> No, it is not a localization.
> Not only is the duration of a UTC second equal to the duration of a TAI
> second, but a UTC clock also ticks its seconds exactly when a TAI clock
> ticks its seconds.  On the other hand, UTC noon is never more than 0.9
> seconds different from the mean noon at Greenwich.  Thus, UTC keeps
> aligned with the gradually slowing of the earth's rotation while TAI
> ignores the earth's rotation.

And timezones do the same thing. They make the time correspond (more
roughly than .9 sec) to the local solar noon. 
The second was defined as 1/86400 of the mean solar day.  It was cast
adrift from that once clocks became sufficiently accurate, since doing
that was easier then remanufacturing all of those clocks. That means
that the second is no longer tied to the sun, even the mean sun rather
than the real sun ( for which the second varies by about .04 percent--
that was the first divorce of the second from the sun)
Thus one could, as Hasler does, aruge that UTC is simply an aberation,
like leap years, and like time zones, from the orderly progression of
the seconds (TAI), and should be handled in exactly the same way as leap
days, daylight saving time, or timezones. Others, apparenly like you,
like the idea that the seconds track the day as well as possible. It is
a matter of preference, not some sort of "natural order". 

> The UTC alignment requires an occasional extra second (always added at
> midnight).  Thus, you might see a UTC clock reporting
> 	23:59:57
> 	23:59:58
> 	23:59:59
> 	23:59:60
> 	00:00:00
> 	00:00:01
> 	00:00:02
> That is, the last minute of a day sometimes (but rarely) has 61 seconds.
>  For systems and computations that require uniform time-intervals
> (delta-t), this is not acceptable.  Those systems and computations rely
> on TAI, which is indeed uniform.  A leap-second file allows conversions
> between UTC and TAI.

More information about the questions mailing list